RESULTS OF THE TESTS ON CONCRETE (PART 1)

Transcription

1 RESULTS OF THE TESTS ON CONCRETE (PART 1) FORGE Report D3.6 VER 1.0 Name Organisation Signature Date Compiled M.V. Villar, P.L. Martín & F.J. Romero CIEMAT M.V. Villar, P.L. Martín & F.J. Romero March 2010 Verified Approved Richard Shaw BGS 23 April 2010 Euratom 7 th Framework Programme Project: FORGE

2 FORGE Report: Dx.xxD3.6 Ver 1.0 Fate of repository gases (FORGE) The multiple barrier concept is the cornerstone of all proposed schemes for underground disposal of radioactive wastes. The concept invokes a series of barriers, both engineered and natural, between the waste and the surface. Achieving this concept is the primary objective of all disposal programmes, from site appraisal and characterisation to repository design and construction. However, the performance of the repository as a whole (waste, buffer, engineering disturbed zone, host rock), and in particular its gas transport properties, are still poorly understood. Issues still to be adequately examined that relate to understanding basic processes include: dilational versus viscocapillary flow mechanisms; long-term integrity of seals, in particular gas flow along contacts; role of the EDZ as a conduit for preferential flow; laboratory to field up-scaling. Understanding gas generation and migration is thus vital in the quantitative assessment of repositories and is the focus of the research in this integrated, multi-disciplinary project. The FORGE project is a pan-european project with links to international radioactive waste management organisations, regulators and academia, specifically designed to tackle the key research issues associated with the generation and movement of repository gasses. Of particular importance are the longterm performance of bentonite buffers, plastic clays, indurated mudrocks and crystalline formations. Further experimental data are required to reduce uncertainty relating to the quantitative treatment of gas in performance assessment. FORGE will address these issues through a series of laboratory and field-scale experiments, including the development of new methods for up-scaling allowing the optimisation of concepts through detailed scenario analysis. The FORGE partners are committed to training and CPD through a broad portfolio of training opportunities and initiatives which form a significant part of the project. Further details on the FORGE project and its outcomes can be accessed at i

3 CIEMAT/DMA/2G207/02/10 FORGE DELIVERABLE 3.6 RESULTS OF THE TESTS ON CONCRETE (PART 1) M.V. Villar, P.L. Martín & F.J. Romero March 2010

4

5 FORGE: RESULTS OF THE TESTS ON CONCRETE (PART 1) 1 INTRODUCTION GAS PERMEABILITY Methodology Non-steady state method Steady-state method Results Non-steady state method Steady-state method UNSATURATED FLOW Methodology Results Column B Column A Column C SUMMARY AND CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES... 30

6

7 FORGE: RESULTS OF THE TESTS ON CONCRETE (PART 1) 1 INTRODUCTION In most countries the final disposal of low and intermediate-level radioactive waste is performed on surface or near-surface facilities, in which concrete is frequently used as a barrier. This work is a contribution to the understanding of the behaviour of concrete barriers in surface disposal facilities, in particular in the Spanish disposal facility of El Cabril, where the waste containers are placed inside concrete cells. The durability of concrete and its mechanical properties are intrinsically bound to moisture transport effects, especially when it is subjected to repeated wetting and drying regimes, and that is why a detailed thermohydraulic characterisation is necessary to model its behaviour. Thus, with a view on LLW and ILW issues, gas permeability measurements have been performed in concrete samples of different degrees of saturation under low gas pressure. In addition, to analyse the unsaturated flow, columns of concrete instrumented with miniature RH/T transmitters have been infiltrated or desiccated. The back-analysis of these experiments allows to compute the relative permeability. The methodologies followed are described in detail in Deliverable 3.3 (Villar et al. 2010). The concrete used was manufactured at the C.A. El Cabril in September 2006 and May 2008 in the form of cylindrical blocks casted in PVC molds of different sizes, following the procedures used to manufacture the disposal cells. The samples were cured at El Cabril at ambient temperature, and once at CIEMAT they were kept in a 100-percent RH atmosphere and at room temperature (Villar et al. 2009). The concrete has a characteristic strength of 350 kp/cm 2, with a water/cement ratio of 0.43 and a consistency of 14 cm. Its average pore size is 0.03 µm (Zuloaga 2008). The specific weight of solid particles of the ground concrete is The concrete theoretical composition, which includes OPC, is the following: aggregates 4/16 sand 0/4 sand 0/2 cement I 42,5R/SR Melcret-222 additive water 1023 kg 634 kg 203 kg 400 kg 6.5 kg 175 L 2 GAS PERMEABILITY Two different experimental setups have been used, a steady and a non-steady state one. The non-steady state method has been used at CIEMAT for more than a decade, whereas the steady state method has been fine tuned for this project (Villar et al. 2010). 1

8 2.1 Methodology Non-steady state method The cartoon of the assembly is shown in Figure 1. The cylindrical concrete sample is placed in a triaxial cell confined between two porous stones and wrapped in two latex membranes, between which vacuum grease is applied in order to prevent the loss of gas. The cell walls are made out of methacrylate and are capable of withstanding pressures up to 3000 kpa. The cell has four inlets drilled in the base, one for the sample top drainage/back pressure, two for the sample bottom drainage/pore pressure, and one for the confining pressure. A pressure is applied to the chamber of the triaxial cell, high enough to ensure perfect adherence of the latex membranes to the walls of the sample (this pressure may have an influence on the permeability value obtained, and has been usually of 0.6 MPa). The inlet at the lower part of the sample is connected to an airtight tank of known volume, in which nitrogen gas is previously injected at a pressure slightly higher than atmospheric. The tank is instrumented with a pressure sensor, connected to a data acquisition system, which records the pressure of the fluid contained inside. The inlet at the upper part of the sample is left open to the atmosphere. The test consists in allowing the air in the tank to exit to the atmosphere across the specimen, while the decrease in pressure in the tank is measured versus time. The test must be performed at a constant temperature. DATA ACQUISITION SYSTEM OUTLET TO ATMOSPHERE PRESSURE TRANSDUCER N 2 TRIAXIAL CELL GAS TANK CONSTANT PRESSURE SYSTEM Figure 1: Schematic representation of the permeability to gas measuring system The permeability to gas is calculated in accordance with the following equation (Yoshimi & Osterberg 1963): k g = 2.3 V L? g Log P 0 ( t) 10 g 0 [1] A P atm P where k g is the permeability to gas (m/s), V the volume of the tank (m 3 ), L the length of the sample (m), A the surface area of this sample (m 2 ), ρ g the density of the gas (kg/m 3 ), P atm is t-t P 2

9 atmospheric pressure (N/m 2 ), P 0 is the excess of pressure over atmospheric pressure in time t 0 (s) and P(t) is the excess over atmospheric pressure in the tank in time t. This equation was developed in a manner analogous to that used for the expression of permeability to water using a falling head permeameter, with the air continuity equation being applied through consideration of compressibility (Lloret 1982). In developing the equation, it has been assumed that the initial P 0 pressures are relatively small compared to atmospheric pressure. Also it must be assumed that while the pressure is decreasing in the tank, the distribution of pressure in the soil sample is the same as would exist if this instantaneous pressure had been maintained in the tank for a long period of time. The volume of the spherical tank used is m 3, the gas used for the tests is nitrogen, for which a density of 1.12 kg/m 3 was taken. The pressure of the tank on test initiation is established at values close to 1.03 bar, since keeping the properties of the gas constant throughout the test requires that it not be subjected to high pressures. Taking into account the dynamic viscosity of nitrogen (µ g, Pa s), the following relation between permeability to gas (k g, m/s) and the product of intrinsic permeability measured with nitrogen gas (k ig, m 2 ) by the relative permeability to gas (k rg ) is obtained: ρ g g 5 k g = kig krg = kig k µ g rg [2] Steady-state method The setup has been designed to perform steady gas permeability measurements under different gas pressures (Figure 2). The cylindrical sample is confined in a triaxial cell and mounted as described above. The cell is filled with water and pressurised with nitrogen, which is separated from the water in the cell through an elastic membrane contained in an OLAER s pressure accumulator (able to withstand pressures of up to 330 bar). The injection and downstream pressures can be independently varied and kept constant during the period of time necessary to get steady flow by HI-TEC gas forward pressure controllers. Associated to the pressure controllers, DRUCK pressure transmitters, PTX1400 series, have been placed at several points. Different range HITECH gas mass flowmeters measure the inward and outward flows (0.2-10, and STP cm 3 /min). The technical details of the equipment are given in Villar et al. (2010). The equipment works like a constant head permeameter, with the possibility to change the head value and measure the gas flow value. The system applies the pressures to the sample and registers flow and pressures from the measurement devices. In and outflow gas rates, up and downstream pressure, temperatures and the confining pressure are monitored. To compute the permeability (intrinsic permeability measured with gas flow, k ig, relative permeability to gas, k rg ) the inflow or outflow measurements can be used, applying the following equation for incompressible media with compressible pore fluids (Scheidegger 1974): k ig k rg Q µ g L 2P = A ( P P ) 2 up 2 dw dw [3] where Q is the flow (volume of fluid as a function of time), A is the sample surface area, µ g is the fluid dynamic viscosity, L is the sample length and P up and P dw are the upstream and 3

10 downstream pressures applied at the bottom (inlet) and the top (outlet), respectively, of the sample. In turn gas permeability, k g, can be computed as in Equation 2, taking into account the gas density and viscosity change with upstream or downstream pressures (P): ρ g g P k g = kig k µ g rg [4] 2.2 Results Figure 2: Schematic diagram of the setup for the gas permeability tests Non-steady state method The specimens used had diameters of 35 or 50 mm and heights between 40 and 70 mm. The same specimen was measured several times after taking it out of the wet chamber, letting it dry at laboratory conditions between the different measurements. In this way the change of gas permeability during drying was evaluated. The gas permeability of some specimens was measured after they were water saturated. Table I and the following ones summarise the results obtained for the measurements performed in each specimen, expressed as gas permeability (k g ) and as intrinsic permeability times relative permeability (k ig k rg ). The effective degree of saturation is also included, e(1-s r ). Except for samples HOR2 and HOR4, the water content and degree of saturation values have to be confirmed once the samples are dried in the oven at the end of the measuring process. Gas permeability values as a function of water content are shown in Figure 3, except for those corresponding to samples HOR9 and HOR10, because no steady permeability values were reached in any of these samples. The permeability increase on drying seems more acute in samples that were previously saturated, which in addition have lower permeabilities. Gas permeability of concrete is between and m 2 for water contents between 2.0 and 6.4 %, corresponding to degrees of saturation between 34 and 96 percent. 4

11 Table I: Results of gas permeability measurements during air drying in sample HOR1 (ρ d =2.22 g/cm 3, e=0.21) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) a a The sample got accidentally wet at the end of the first measurement and it was dried at 60 C Table II: Results of gas permeability measurements in sample HOR2 during air drying before and after saturation (ρ d =2.32 g/cm 3, e=0.16) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) After saturation (HOR2s) Table III: Results of gas permeability measurements in sample HOR3 after saturation and air drying (ρ d =2.30 g/cm 3, e=0.16) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) a a

12 T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) a a The sample got accidentally wet Table IV: Results of the gas permeability measurements in sample HOR4 after saturation and air drying and then during air drying after resaturation (ρ d =2.33 g/cm 3, e=0.15) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) After resaturation (HOR4_2) Table V: Results of the gas permeability measurements in samples HOR5 during air drying after saturation and resaturation (ρ d =2.29 g/cm 3, e=0.17) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) After resaturation (HOR5_2) Table VI: Results of gas permeability measurements in sample HOR6 during air drying after saturation (ρ d =2.26 g/cm 3, e=0.19) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s)

13 Table VII: Results of gas permeability measurements in sample HOR7 during air drying (ρ d =2.26 g/cm 3, e=0.19) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) Table VIII: Results of gas permeability measurements in sample HOR9 during air drying (ρ d =2.27 g/cm 3, e=0.18) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) a a a Non stable values Table IX: Results of gas permeability measurements in sample HOR10 during air drying (ρ d =2.28 g/cm 3, e=0.18) T ( C) w final (%) S r final (%) k ig k rg (m 2 ) e (1-S r ) k g (m/s) a a Non stable values Intrinsic permeability depends just on the material characteristics, and its value can be obtained either with a liquid in the saturated material or with gas in the completely dried material, provided that the fluid does not interact with the material. However, if the fluid modifies the characteristics of the material, the value of intrinsic permeability obtained could be altered. The gas permeability measurements performed up to now did not allow to derive the intrinsic permeability, since the specimens were not completely dried. Drying the samples in the oven up to zero water content would have implied modifying their microstructure, and that is why they were just air dried to get different water content values. An indirect method to derive intrinsic permeability from gas permeability measurements was presented in Villar (2000) and Villar & Lloret (2001), and has been applied to the measurements shown above. Figure 4 shows the gas permeability change with water content in samples that had not been previously saturated (HOR1, HOR2 y HOR7). This parameter presents a better correlation with the permeability values obtained than accessible pore volume or degree of saturation. The following relationship between gas permeability (k g, m/s) and water content (w, %) can be deduced: k g = w [5] If in this equation water content is taken as 0, the gas permeability of the dried concrete would be obtained (k g0, m/s), which cannot be obtained through direct measurement because of the impossibility of drying completely the samples without modifying their microstructure. Then, the intrinsic permeability of concrete measured with air flow could be obtained through Equation 2, considering k rg equal to 1, since the degree of saturation would be 0. The value thus obtained is m 2, which would correspond to concrete not previously saturated. It 7

14 is in the order of those values obtained in saturated samples with water flow (Villar et al. 2009). Gas permeability (m/s) 1.0E E E E-11 HOR1 HOR2 HOR2s HOR3 HOR4 HOR4_2 HOR5 HOR5_2 HOR6 HOR7 1.0E-12 bold symbols: specimens not previously saturated Water content (%) Figure 3: Concrete gas permeability evolution during air drying 1.0E-09 HOR1, HOR2, HOR7 Gas permeability (m/s) 1.0E Water content (%) Figure 4: Gas permeability as a function of water content in samples not previously saturated (preliminary results) 8

15 2.2.2 Steady-state method Some of the samples measured with the non-steady state method were also measured in the steady state equipment. Sample HOR3 was measured twice during the drying process. The sample water content in the first measurement (HOR3_6, Table X) was 4.0 percent (to be confirmed). Between this measurement and the next one in the steady state equipment (HOR3_7, Table XI), the sample was accidentally wetted, and that is why the water content in the latter measurement was higher (presumably 4.7 percent). During the tests, the confining pressure was kept constant (at 4-5 bar in the first test and 15 bar in the second one) and the backpressure was kept atmospheric. The injection pressure was first gradually increased and then reduced. The results obtained in this phase for the two tests are shown in Figure 5. As expected, the sample with the higher water content showed the lower gas permeability. The values obtained during the pressure increase were equal to those found on pressure reduction. Permeability decreased slightly at the beginning of the injection pressure increase cycle. Also, the k ig k rg values calculated with the inflow or the outflow for a given pressure condition are equal, whereas the k g calculated with the inflow is not the same than that calculated with the outflow, since the gas density in both cases is different due to the pressure difference between top and bottom of the sample. Figure 5 shows also the permeability determined in the non-steady state equipment in the sample with the same water content. In both cases that value is higher, what could be due to the very low injection pressure in the tests performed in the non-steady state equipment. In test HOR3_7, after the pressure cycle under constant confining pressure, the confining pressure and the injection pressure were reduced to 5 and 2 bar, respectively, and then increased gradually and simultaneously to 15 and 10 bar, respectively. Table X: Results of gas permeability measurements in sample HOR3 (test HOR3_6) obtained with the steady-state method (ρ d =2.30 g/cm 3, w=4.0%) Duration (days) Confining P (bar) Injection P (kpa) Back P (kpa) k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s)

16 Table XI: Results of gas permeability measurements in sample HOR3 (test HOR3_7) obtained with the steady-state method (ρ d =2.30 g/cm 3, w=4.7%) Duration (h) Confining P (bar) Injection P (bar) Back P (bar) k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s)

17 Gas permeability (m/s) 1.0E E E E-12 BackP atm Conf P=15 bar, w=4.7% Conf P=4-5 bar, w=4.0% Hydraulic head (bar) Figure 5: Change of gas permeability (calculated from outflow) with injection pressure for sample HOR3 at different water contents and constant confining pressures. Atmospheric backpressure. The values obtained in the non-steady state equipment are shown with dotted horizontal lines Sample HOR5 was measured twice during the air drying process, with two different water contents, 3.4 and 2.3 percent (to be confirmed). A summary of the values obtained during the first measurement is shown in Table XII (HOR5_6). In test HOR5_6 the backpressure was initially kept constant at 174 kpa while the injection pressure was increased under a constant confining pressure of 7 bar (Phase 1); afterwards, the backpressure was increased and then decreased while injection pressure was kept constant at 800 kpa and the confining pressure at 10 bar (Phase 2); and finally, keeping a constant backpressure of 2 bar and a confining pressure of 10 bar, the injection pressure was reduced (Phase 3). After each pressure change flow became quickly stable and for this reason the duration of each step was usually of just a few hours. The permeability values obtained during the test calculated from the gas outflow are plotted in Figure 6 as a function of hydraulic head. All the values are lower than those determined with the non-steady state method for the same water content. While injection pressure was kept constant at 800 kpa (Phase 2), permeability decreased as hydraulic head increased (i.e. as backpressure decreased). While backpressure was kept constant and injection pressure varied, permeability did not show a clear dependence on hydraulic head. For low hydraulic heads, there was a significant difference between the gas permeability values obtained for low and high backpressures, those obtained for high backpressures being higher. Thus, the determining factor for the permeability increase seems to be the backpressure, which is the one predominant inside the sample. 11

18 Table XII: Results of gas permeability measurements in sample HOR5 (test HOR5_6) obtained with the steady-state method (ρ d =2.29 g/cm 3, w=3.4%) Duration (days) Confining P (bar) Injection P (kpa) Back P (kpa) Phase 1 k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s) Phase < < < < < < Phase

19 Permeability (m/s) 1.0E E-11 Phase 1: bp 174 kpa Phase 2: Inj.P 800 kpa Phase 3: bp 200 kpa 1.0E Hydraulic head (kpa) Figure 6: Gas permeability calculated from outflow during test HOR5_6 (ρ d =2.29 g/cm 3, w=3.4%). Confining pressure was 7 bar in Phase 1 and 10 bar in Phase 2 and 3. The thick horizontal line is the value determined in the non-steady state equipment. bp: backpressure, Inj.P: injection pressure The results obtained during the second measurement, performed with a water content of 2.3 percent (to be confirmed), are shown in Table XIII (HOR5_8). There were two phases in this test: in Phase 1, under a constant confining pressure of 15 bar and atmospheric backpressure, the injection pressure was gradually increased and then decreased; during Phase 2, both confining pressure and injection pressure were increased and later decreased simultaneously, so that to keep an approximately constant deviator between 5 and 8 bar. The values obtained during Phase 1 as computed from the gas inflow and outflow are plotted in Figure 7, together with the value obtained in the non-steady state method. The difference between the values obtained from the inflow and the outflow, which is significant, came from the change of gas density with pressure. Since backpressure was kept atmospheric, the values calculated from the outflow were not affected by the gas density change, although they showed a dependence on injection pressure for the low hydraulic heads (permeability decreased with injection pressure increase up to 4 bar hydraulic head). The values obtained during the injection pressure increase and decrease were the same for the same hydraulic head values. During Phase 2, permeability decreased as the confining pressure was higher. All the values were lower than the value obtained in the non-steady state equipment for the same water content. Besides, the values were lower than those determined in test HOR5_6, despite the fact that the water content in the latter was higher. This could be due to the different experimental conditions (pressures, in particular the higher confining pressure), to which gas permeability is very sensitive, or to the accidental wetting of the concrete during test HOR5_8, whose final water content was not checked. That would also explain the much lower permeability determined in the steady state equipment than in the non-steady state one. 13

20 Table XIII: Results of gas permeability measurements in sample HOR5 (test HOR5_8) obtained with the steady-state method (ρ d =2.29 g/cm 3, w=2.3%) Duration (h) Confining P (bar) Injection P (bar) Back P (bar) Phase 1 k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s)

21 Duration (h) Confining P (bar) Injection P (bar) Back P (bar) k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s) Phase E-09 Permeability (m/s) 1.0E E-11 inflow outflow 1.0E Hydraulic head (bar) Figure 7: Gas permeability calculated from inflow and outflow during Phase 1 of test HOR5_8 (ρ d =2.29 g/cm 3, w=2.3%). Confining pressure was 15 bar and atmospheric backpressure, injection pressure was increased and decreased. The thick horizontal line is the value determined in the nonsteady state equipment Sample HOR7 was measured with a water content of 1.6 percent (to be confirmed) in two phases, in both of them the backpressure was kept atmospheric. In Phase 1 the injection and the confining pressures were simultaneously increased, so that to keep a constant deviator of 3 bar. In Phase 2 the confining pressure was kept constant (14-16 bar) and the injection pressure 15

22 was gradually decreased. The results obtained in both phases are summarised in Table XIV and plotted in Figure 8. Overall gas permeability decreased with confining pressure and hydraulic head. Table XIV: Results of gas permeability measurements in sample HOR7 (test HOR7_3) obtained with the steady-state method (ρ d =2.26 g/cm 3, w=1.6%) Confining P (bar) Injection P (bar) Back P (bar) k ig k rg inflow (m 2 ) Phase 1 k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s) Phase

23 1.0E-11 Gas permeability (m/s) Phase 1 Phase 2 1.0E Hydraulic head (bar) Figure 8: Gas permeability calculated from outflow during test HOR7_3 (ρ d =2.26 g/cm 3, w=1.6%). Atmospheric backpressure. Phase 1: confining and injection pressures increased simultaneously, Phase 2: confining pressure at bar, injection pressure decreased Samples HOR9 and HOR10 were tested immediately after they were taken out of the wet chamber, however they showed high permeability, both when tested with the non-steady state method (Table VIII and Table IX) and with the steady state one. The results obtained in the steady state equipment are summarised in Table XV and Table XVI. Both samples were tested initially under a confining pressure of 15 bar and atmospheric backpressure while the injection pressure was gradually decreased. The results obtained during this phase in both samples are plotted in Figure 9. The gas permeability kept constant for sample HOR9 despite the injection pressure change, while it decreased for sample HOR10 as injection pressure increased. Sample HOR10 was not further tested, since the gas flow was so high that the pressure steps could not be kept for long; but sample HOR9 was subjected to a second testing phase in which the confining and the injection pressures were simultaneously decreased (Figure 10). As observed in other tests, the gas permeability increased as confining pressure decreased, regardless of the injection pressure applied. 17

24 Table XV: Results of gas permeability measurements in sample HOR9 (test HOR9_2) obtained with the steady-state method (ρ d =2.27 g/cm 3, w=3.4%) Duration (h) Confining P (bar) Injection P (bar) Back P (bar) Phase 1 k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s) Phase Table XVI: Results of gas permeability measurements in sample HOR10 (test HOR10_1) obtained with the steady-state method (ρ d =2.28 g/cm 3, w=4.0%) Duration (h) Confining P (bar) Injection P (bar) Back P (bar) k ig k rg inflow (m 2 ) k ig k rg outflow (m 2 ) k g inflow (m/s) k g outflow (m/s)

25 1.0E-10 Permeability (m/s) 1.0E-11 HOR9 HOR10 1.0E Hydraulic head (bar) Figure 9: Gas permeability calculated from outflow for samples HOR9 (ρ d =2.27 g/cm 3, w=3.4%) and HOR10 (ρ d =2.28 g/cm 3, w=4.0%). Atmospheric backpressure, confining pressure 15 bar 1.0E-09 Permeability (m/s) 1.0E E Confining pressure (bar) Figure 10: Gas permeability calculated from outflow for sample HOR9 in Phase 2 of test HOR9_2 (ρ d =2.27 g/cm 3, w=3.4%). Atmospheric backpressure, confining and injection pressures decreased simultaneously 19

26 3 UNSATURATED FLOW 3.1 Methodology The unsaturated flow in concrete can be evaluated through the determination of the unsaturated permeability, which can be accomplished by the back-analysis of evaporation tests in which the proper variables are measured: total water change and suction at different levels. The values of suction measured can be converted to water content through the retention curve. This method is a modification of the instantaneous profile method. The concrete drying that takes place in any structure at ambient conditions implies water movement to the drying surface where it escapes as vapour into the surrounding air. The water lost from the surface by evaporation may be replaced by water flowing under the action of capillary forces from the interior to the surface, or the evaporating surface may fall below the surface of the solid, and water vapour must then traverse a certain length of pore space before it can escape at the surface (Schaffer 1932). Any of these movements gives place to variable space and time water content gradients. The concrete water content conditions its mechanical properties, due to the interaction between the porous solid and the interstitial water, which in turn is related to the concrete shrinkage and other weathering processes. The drying process generally includes unsaturated flow of liquid within the porous solid, vapour flow within the porous solid, the liquid-vapour phase change, and convective-diffusive transfer of vapour from the surface of the solid to the surroundings. For these reasons, the evaporation tests performed at CIEMAT can be used to calculate relative permeability by back-analysis. The setup for the evaporation tests is very simple (Figure 11). Lab RH s1: 1 cm 8.5 cm s2: 4 cm s3: 8 cm 5 capacitive RH sensors s5: 18 cm s4: 13 cm 20 cm Balance Figure 11: Setup for the evaporation tests in concrete The columns used for the evaporation tests were prepared in July 2008 at El Cabril facilities in PVC moulds of internal diameter 8.5 cm and height 20 cm. Five 0.8-cm diameter perforations had been previously drilled at different levels along each column, and plastic dummy cylinders had been placed in them up to a depth of 5 cm. Once the concrete cast, these cylinders were replaced by sensors. The sensors used were Sensirion SHT75 relative humidity 20

27 (RH) and temperature (T) sensors. These sensors have an accuracy of 2 percent RH in the range between 20 and 80 percent RH. An additional sensor was placed 12 cm above the column in order to record the ambient changes. The measurements of the sensors were converted to suction through the psycrometric law. The upper surface of the column was open to atmosphere so that water could freely evaporate. The column was placed over a balance to record mass changes during the evaporation process. Both the sensors readings and the column mass were periodically recorded by a data acquisition system specifically designed. Additionally, after the evaporation test, an infiltration phase could be performed by adding water to the top surface of the column. At the end of the test the columns were cut into five cylindrical sections whose dimensions were measured and the water content of them was determined by oven drying at 110 C. 3.2 Results The laboratory facilities allow simultaneous testing of two columns. Column A and Column B were concurrently tested, but Column A was subjected to infiltration after the evaporation phase, whereas Column B was dismounted after evaporation. Column C was later tested and dismounted after evaporation. All these columns had by mistake the lid placed on the surface intended for evaporation, and for this reason evaporation took place initially through the bottom of the column for different periods of time. Before the beginning of the instrumented evaporation test, the lids were changed to the bottom of the columns. This explains the difference in the relative humidities along the columns measured at the beginning of the evaporation tests Column B Column B was manufactured on 29/5/08 and had the lid placed on top up to 24/7/08, what means that for two months evaporation took place through the wrong end. For this reason the relative humidities measured at the beginning of the evaporation test were not uniform, but lower towards the bottom. The evaporation instrumented test went on for 306 days, and Figure 12 shows the suctions measured inside the column computed from the sensors measurements. The sensor closest to the evaporation surface (s1) recorded a sharp increase of suction, the next one (s2) recorded a softer increase, and the other sensors recorded an initial increase that seemed to stabilise after 4000 h of testing. Sensor s4 failed before the end of the test. The fact that the bottom of the column was open to atmosphere for 2 months was reflected in the high initial suction recorded by sensor s5. The initial suction increase recorded by all sensors (RH decrease) could be due to the continuous hydration of cement that takes place while the RH in the pores is higher than 85 percent, through which this water is linked to the concrete microstructure (Selih & Bremmer 1996). At the beginning of the test all sensors recorded RH higher than 84 percent, and consequently it is possible that the casting process was still going on. A sensor was placed above the column 1683 h after the beginning of the test, and recorded an average temperature during the test of 23 C and a relative humidity of 38 percent. The temperature inside the column was uniform and equal to that in the laboratory, which reflected the daily and seasonal changes (Figure 13). The mass change measured during the test is also shown in this figure, where an intense initial evaporation followed by a more gradual one after approximately 1000 h can be observed. The water loss was sharper again during the summer, approximately 5000 h after the beginning of the test. 21

28 Suction inside concrete (MPa) Laboratory suction (MPa) Evaporation time (h) s1 s2 s3 s4 s5 s amb Figure 12: Evolution of suction computed from the sensors measurements during evaporation of Column B (s1 at 1 cm from the column surface, s2 at 4 cm, s3 at 8 cm, s4 at 13 cm, s5 at18 cm, s amb: laboratory suction) mass temperature 28 Water lost (g) Temperature ( C) Evaporation time (h) Figure 13: Mass change of the column and laboratory temperature during evaporation of Column B 22

29 Figure 14: Appearance of the sections cut from Column B Final water content (%) water content relative humidity Final RH (%) Distance from evaporation surface (cm) Figure 15: Gravimetric water content measured at the end of test Column B and relative humidity recorded by sensors inside concrete just before dismantling Column A Column A was manufactured on 29/5/08 and had the lid placed on top up to 23/7/08, what means that for two months evaporation took place through the wrong end. For this reason the relative humidities measured at the beginning of the evaporation test were not uniform, but lower towards the bottom (higher suctions, particularly in sensors s3, s4 and s5). The evaporation went on for 343 days. Figure 16 shows the suctions measured inside the column during the evaporation test computed from the sensors measurements. The sensor closest to the evaporation surface (s1) recorded a sharp initial increase of suction, followed by a softer but steady increase. Sensors s2, s3 and s4 recorded also a sharper suction increase for 600 h that was later attenuated. As mentioned for Column B, the initial suction increase recorded by all sensors (RH decrease) could be due to the continuous hydration of cement that takes place while the RH in the pores is higher than 85 percent, since at the beginning of the test all sensors recorded RH higher than 90 percent. After 4000 h of testing, sensors s3 and s4 seemed to record a steady RH value, while s2 (at 4 cm from the evaporation surface) continued recording a soft suction increase. Approximately after 6000 h of testing there is an overall slight increase in suction all along the column due to the seasonal temperature increase. The fact that the bottom of the column was open to atmosphere for 2 months was reflected in the high initial suction recorded by sensor s5. The continuous increase in suction recorded by this sensor could be due to a bad sealing of the perforation. A sensor was placed above the column 1803 h after the beginning of the test, and recorded an average temperature during the test of 23 C and a relative humidity of 38 percent. The 23

30 temperature inside the column was uniform and equal to that in the laboratory, which reflected the daily and seasonal changes (Figure 17). The mass change measured during the test is also shown in this Figure, where an intense initial evaporation followed by a more gradual one after approximately 1000 h can be observed. The water loss was sharper again during the summer, approximately 5000 h after the beginning of the test Suction inside concrete (MPa) Laboratory suction (MPa) Time (hours) s1 s2 s3 s4 s5 s lab Figure 16: Evolution of suction computed from the sensors measurements during evaporation of Column A (s1 at 1 cm from the column surface, s2 at 4 cm, s3 at 8 cm, s4 at 13 cm, s5 at18 cm, s lab: laboratory suction) After the evaporation phase an infiltration phase was carried out. The PVC column was prolonged and a sponge was placed on the upper surface of the concrete column (Figure 18). This sponge was periodically moistened with deionised water. To reduce evaporation the upper part of the cell and sponge were covered with paraffin foil. Although the mass column was recorded, the setup did not allow to separate the mass changes due to concrete water intake from those due to the sponge wetting or evaporation, and consequently, these measurements were not taken into account. After 100 h hydration was stopped to solve some leakages and the top of the column was exposed to air drying for 48 h before taking up hydration. Figure 19 shows the evolution of suction inside the concrete during infiltration. All the sensors, except s4, recorded suction decrease during hydration. The sensors placed in the column ends (s1 and s5) immediately reflected the beginning of hydration by a sharp suction decrease. The fact that the sensor located farthest from the hydration surface (s5) recorded quickly the RH increase means that there was a water path other than the concrete matrix. Probably during drying some shrinkage took place and the concrete column could have slightly separated from the PVC wall. The faulty sealing of sensor s5 could have contributed to the arrival of water to the position where it was located. The negative suction values are 24

31 due to RH measurements higher than 100 percent caused by the flooding of the sensors, which is also the reason for s3 and s4 stopping working. The temperatures inside the concrete were homogeneous, but 0.4 C lower than those in the laboratory mass temperature Water lost (g) Evaporation time (h) Temperature ( C) Figure 17: Mass change of the column and laboratory temperature during evaporation of Column A Figure 18: Appearance of Column A during the hydration phase 25

32 Suction inside concrete (MPa) Laboratory suction (MPa) Tiempo (h) s1 s2 s3 s4 s5 s lab Figure 19: Evolution of suction computed from the sensors measurements during infiltration of Column A (s1 at 1 cm from the column surface, s2 at 4 cm, s3 at 8 cm, s4 at 13 cm, s5 at18 cm, s lab: laboratory suction) Figure 20: Wet cutting of the concrete column at the end of the test Column A was dismounted after 343 days of drying and 109 days of hydration. The column was wet cut with a diamond disc in five transversal sections, each sensor being centred in one of the sections (Figure 20). The gravimetric water content values determined in these sections are shown in Figure 21, along with the dry density values calculated taking into account the dimensions of each section. Surprisingly, the final water content was higher towards the 26

33 bottom of the column while the dry density decreased in the same direction. This could be due to irreversible shrinkage during drying, which would have affected mainly the upper half of the column, causing a density increase that did not change during hydration, which was much quicker, and limited the quantity of water the concrete was able to absorb. Another explanation for this water content and density distribution could be that hydration took place mainly through the sensor drillings and the external wall, instead of through the concrete matrix. The average final water content was 5.4 percent and dry density 2.21 g/cm 3. There is uncertainty about the initial values of the column, but the best estimation points to an initial water content of 5 percent. Water content (%) water content density Distance from surface (cm) Dry density (g/cm 3 ) Figure 21: Concrete gravimetric water content and dry density measured at the end of test Column A Column C Column C was manufactured on 29/5/08 and placed in a wet chamber in July 2008 with the cover placed on top. It means that for almost 2 months evaporation took place through the wrong end, although the humidity loss could be recovered in the wet chamber, where the column with the cover on top stood for 8 months. In April 2009 the cover was removed and fixed at the bottom of the column. The column was finally taken out of the wet chamber in 9/6/09 for the evaporation test. The initial dry density and water content of the column, checked at the end of the test, were 2.20 g/cm 3 and 4.7 percent, with an initial mass of 2572 g, corresponding to a dry mass of 2455 g. The relative humidities measured at the beginning of the evaporation test were quite uniform, around percent. Figure 22 shows the suctions measured inside the column during the evaporation test computed from the sensors measurements. There were some problems with the sensor placed closest to the evaporation surface and its recordings started some hours after the initiation of the test. The sharpest suction increase was recorded by sensor s1, followed by sensor s2. The sensor at the bottom of the column (s5) recorded also a significant increase. The sensor placed above the column recorded an average temperature during the test of 26 C and a relative humidity of 35 percent. The temperature inside the column was uniform and equal to that in the laboratory, which reflected the daily and seasonal changes (Figure 23). The mass change measured during the test is also shown in this figure, where an intense initial evaporation followed by a more gradual one after approximately 1800 h can be observed. The initial period coincided with the high summer temperatures and low relative humidities. 27

34 Suction inside concrete (MPa) Laboratory suction ( C) Time (hours) s1 s2 s3 s4 s5 s amb Figure 22: Evolution of suction computed from the sensors measurements during evaporation of Column C (s1 at 1 cm from the column surface, s2 at 4 cm, s3 at 8 cm, s4 at 13 cm, s5 at18 cm, s amb: laboratory suction) Water loss (g) corrected mass T Temperature ( C) Time (h) Figure 23: Mass change of the column and laboratory temperature during evaporation of Column C 28

35 The test was dismantled after 252 days of evaporation, the final mass of the concrete column being 2553 g, implying a water loss of 18.8 g, approximately in the order of that measured by the balance during the test. The column was cut with a diamond disc in 5 transversal sections, each sensor being centred in one of the sections. The gravimetric water content values determined in these sections are shown in Figure 25, the change trend along the column being similar to that in RH recorded by the sensors. The water content decreased during the evaporation test all along the column, but particularly in the 6 cm closest to the evaporation surface. The average final water content was 4.0 percent. The final dry densities along the column were between 2.18 and 2.23 g/cm 3, without a particular longitudinal trend. Figure 24: Appearance of the sections cut from Column C Final water content (%) w RH Final relative humidity (%) Distance from surface Figure 25: Gravimetric water content (w) measured at the end of test Column C and relative humidity (RH) recorded by sensors inside concrete just before dismantling 4 SUMMARY AND CONCLUSIONS In order to perform the hydraulic characterisation of concrete, particularly with respect to its gas (vapour) transport properties, two methodologies were fine tuned: A setup for the measurement of gas permeability of concrete with different degrees of saturation under different pressure conditions. 29