- Email: [email protected]
Moisture ingress in rammed earth: Part 1—the effect of soil particle-size distribution on the rate of capillary suction
Construction and Building Materials 18 (2004) 269–280
Moisture ingress in rammed earth: Part 1—the effect of soil particle-size distribution on the rate of capillary suction Matthew Hall*, Youcef Djerbib Centre for the Built Environment, Sheffield Hallam University, Unit 9 Science Park, Howard Street, Sheffield S1 1WB, UK Received 19 November 2003; received in revised form 24 November 2003; accepted 24 November 2003
Abstract The novel initial rate of suction (IRS) ‘wick’ test has been presented, and is suitable for determining the rate of capillary moisture ingress in unstabilised rammed earth that slakes on contact with water. Experimental testing was performed using the ‘wick’ test to investigate the effect of soil particle-size distribution on moisture ingress in rammed earth. Rammed earth generally absorbs much less water due to capillary suction, and at a slower rate, than conventional masonry building materials such as bricks and concrete. Moisture ingress in rammed earth, due to capillary suction, increases linearly per unit inflow surface area against the square root of elapsed time yt . The particle-size distribution of the soil is critical in determining the rate at which moisture may ingress. In a suitable soil, the ratio between the total specific surface area (SSA) of the aggregate fraction and the mass of the binder fraction appears to be positively linked with the rate of capillary suction in rammed earth. Experimental data have been included. 䊚 2003 Elsevier Ltd. All rights reserved.
Ž .
Keywords: Rammed earth; Moisture ingress; Capillary suction
1. Introduction Rammed earth can be defined as ‘«an ancient earth building technique that involves dynamically compacting moist sub-soil between removable shuttering to create an in-situ monolithic compressed earth wall that is both strong and durable’ w1x. Figs. 1 and 2 show a rammed earth dwelling constructed by RAMTEC Pty Ltd in Western Australia, illustrating the current levels of sophistication in design and application of this natural, sustainable material. Some current public perceptions of rammed earth as being a material with low climatic durability are misguided but not entirely unfounded. Rammed earth construction is popular and performs well in warm, dry climates. However, research is essential in determining the application of rammed earth in temperate damp climates if it is to succeed in counties such as the UK. Cavity wall masonry construction, which is by far the most common form in Britain, has largely eliminated the problems of penetrating damp*Corresponding author. Tel.: q44-0-114-225-3200; fax: q44-0114-225-3206. E-mail address: [email protected] (M. Hall).
ness that often plagued the solid walls of older buildings. As rammed earth walls are monolithic, the capillary movement of moisture within them is a particular problem, especially since there are no chemically dosed layers of mortar to suppress it w2x. A variety of chemical additives and admixtures that are available commercially could be used to alleviate the problem but current data relating to their effectiveness is not forthcoming and so poses serious problems for earth building contractors. These problems may be exacerbated depending upon the composition of the soil mix used to form the rammed earth walls; the first logical step is to understand the material before investigating how best to treat it. We have tried many things before regarding admixtures for damp ingress-all too expensive or unreliable«Does adding clay make it better or worse? Does removing clay (e.g. by washing) make it better or worse? Does lessening clay% (e.g. by adding washed sand) make it better or worse? These are real questions w2x.
2. Moisture ingress by capillary suction Excessive moisture content in building elements is referred to as damp w3x. Dampness can enter into and
0950-0618/04/$ - see front matter 䊚 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2003.11.002
270
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 1. Modern rammed earth house in Western Australia (Courtesy RAMTEC Pty Ltd 䉷2001).
move within the fabric of porous construction materials by a variety of different processes. Porous materials are often permeable by air and moisture because they contain a network of open channels. According to the Building Research Establishment w4x, there are two classifications for these channels dependent upon their diameter. A channel with a nominal diameter of 5 mm or above is classed as a pore whereas a channel with a nominal diameter of less than this value is classed as a capillary or a micropore. It should be noted that pores and capillaries are not cylindrical but in fact often exhibit a high level of tortuosity w5x. They may also be interlinked; the extent of which largely determines the permeability of a porous material. Hall w6x observed that if, hypothetically, the solid material of a porous building material is assumed to be inert (i.e. no mass transfer occurs), and that specific boundary effects and dissolved atmospheric gases in the water are ignored a considerable simplification ensues. The role of the solid porous material is then reduced to acting as a pore network that ‘partially constrains the boundary surfaces of the water phases’ to a particular geometry. The system then contains a single component—water—that may exist in any combination of solid, liquid or gaseous form. The specific moisture content w of this idealised porous solid can thus be
Fig. 2. Inside a modern rammed earth home (Courtesy RAMTEC Pty Ltd 䉷2001).
determined, where mw denotes the mass of water and ms the mass of the dry solid: ws
mw . ms
(1)
In practice, the value of w is determined gravimetrically as the difference between the total weights of the sample when it is oven dried to a constant weight at 105 8C, and when it is wet. The conventional description of the reference moisture content of a porous material allows us to clearly distinguish between the hygroscopic moisture content and the capillary moisture content, as can be seen from Fig. 3. The theoretical point of total saturation (Wmax) is rarely achieved, and neither is a completely water-free state (0). In the nominal dry state a degree of hygroscopic water remains, defined by the point Wh, whereas above the nominal point of saturation (Wsat) techniques such as vacuum saturation are required in order to achieve higher levels of moisture content. Movement of moisture in walls is controlled by the masonry and by discontinuities (e.g. fracturing) within the material w5x. In terms of the mechanisms of moisture movement we are dealing with ‘«a single fundamental process, the movement of water through a permeable material whose water content is non-uniform and gen-
Fig. 3. Reference moisture contents, adapted from de Freitas et al. (1996).
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
erally less than saturation’ w6x. To illustrate the process of moisture ingress by capillary suction, Hall w6x used the classic example of a dry brick that has been placed in a shallow tray containing clean water (i.e. the initial rate of suction test). From the moment this is done water is absorbed into the brick by capillary action, and then attempts to distribute itself throughout the pore network of the brick itself. The imbibed water partially displaces the air that previously occupied the pores, whilst at the surfaces that are not actually in contact with water evaporation has begun to occur. At this point, equilibrium can be achieved between evaporative loss and water absorption. However, evaporation entails cooling at the boundary surfaces, which results in a heat flow process being generated inside the brick. The resultant heat gradient modifies the rate of flow of moisture inside the brick. Soluble salts are often dissolved and redeposited at the surface where evaporation is occurring, and the resultant crystalline deposits referred to as efflorescence. As a material approaches saturation, the capillary suction approaches zero w7x. Capillary suction in porous materials, therefore, occurs by the progressive displacement of water so that pockets of air entrapped in the pore and capillary system cannot stop it w8x. The only exception to this occurs where the capillary pressure is simply counteracted by the pressure of entrapped air in a duct that is sealed and has no outlet. Vos and Tammes w9x performed various experiments on the capillary movement of water and concluded that water moving through a porous material by capillarity could travel twice as far in a horizontal direction than in a vertical direction, owing to the effects of gravity. Kieslinger w10x noted that a good correlation between field data and the equation for determining the height of capillary rise had been observed: hs
2s Rd
(2)
where h: water height achieved (cm); s: (constant) 0.074 gycm3; d: density of water (approx. 1 gycm3); R: radius of capillary (cm). Capillary rise is therefore dependent on the diameter of the tube and the liquid within the system. Mamillan w11x gives the following example where, in theory, a 1 mm pore could enable water to rise by 15 mm; a 0.1mm pore enables a capillary rise of 150 mm; and a 0.01 mm pore enables a rise of 1500 mm. The durability of many types of natural building stones has been related to pore size w5x, such that those with high microporosity are often less durable than those with a lower content of micropores. According to Vos and Tammes w12x, if a temperature gradient exists along a capillary then water flow will occur in the direction of lower temperature. This is
271
because surface tension of the capillary wall material is a ‘monotonously decreasing function of temperature’. Conversely, water held within a capillary of uniform temperature but with a gradient in diameter will flow towards the narrower diameter. Consequently, if two capillaries communicate with one another moisture transfer will occur between the material with larger pore diameters to that with the smaller pore diameters until the gradient of unequal surface tension has been resolved through equilibrium w12x. 3. Initial rate of suction (IRS) test The British Standard BS 3921 w13x test for determining capillary suction of water into masonry materials is known as the initial rate of suction (IRS) test. The test method is essentially a gravimetric determination of sorbed water in a partially immersed porous building material (due to capillary suction) over time, the value for which is expressed in kilograms per meter square minute. Hall w14x previously observed the cumulative absorbed mass of water (mw) per unit area of the inflow surface, increases linearly against the square root of elapsed time Žyt.. The sample is to be kept ‘dry’ prior to testing. The current IRS test is, however, of little use for materials that slake (e.g. earth materials) because the determination of imbibed water mass requires the mass of the sample to be kept constant. Furthermore, the test requires the wetted sample surface to be wiped with a damp cloth to remove excess water. This presents a large degree of operator error and a potential drying effect depending upon the type, and moisture content, of the cloth. The testing procedure can be enhanced by maintaining a constant water temperature of 20 8C ("18) in order to ensure constant viscosity. Hall and Kam-Ming Tse w15x recommend that the weighing operation be completed as quickly as possible, ideally within 30 s. Minke w16x has discussed the German standard version of the IRS test, DIN 52 617, within the context of earth building materials. The main differences involve encasing the sample cube sides in fibre-reinforced polyester resin, gluing filter paper to the test face and then placing it on a submerged polyurethane foam base. However, this level of sample intervention is perhaps excessive and deviates significantly from the ‘natural’ conditions of the BS 3921 IRS test. The impermeable resin coating constricts the natural expansion of the sample and restricts the displacement of air from the pore network caused by the imbibed water. Furthermore, equilibrium between evaporative loss at the sample faces and water absorption cannot occur. 4. The IRS ‘wick’ test The IRS ‘wick’ test is a novel adaptation of the current BS 3921 IRS test apparatus (see Fig. 4) devised
272
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 4. BS 3921 IRS test apparatus with IRS ‘wick’ test peripherals.
by the author (Hall). A 30-mm high Oasis娃 ‘wick’, which is 80 mm in diameter, acts as the point of contact for the sample and represents free water being sorbed from another saturated porous medium that offers negligible capillary resistance. However, according to the studies on interface phenomena performed by De Freitas, Abrantes and Crausse w17x, theoretically there should be a small degree of hydric resistance that conditions the maximum flow of moisture transmitted from the wick to the sample face being tested. The contact area is constant and there are no edge effects or meniscus errors on inflow-surface area calculations, as was previously the case in the BS 3921 IRS test. Materials that slake in contact with water are stable throughout the test and negligible mass loss occurs as the inflow surface is retained by the self-weight of the sample acting on the solid surface of the wick. The wick itself can simply be taken out, washed and reused to maintain accuracy and repeatability. Testing was performed on C30 concrete 100=100mm cubes and 3 types of brick; London Brick ‘Fletton’ (high porosity), London Brick ‘Dapple Light’ (medium porosity) and Engineering Brick (low porosity). Six representative samples of each material type were selected. The bricks were cut on a diamond bit saw to give a 100=100-mm inflow-surface area the same as the cube samples. Firstly, the IRS test was performed in accordance with BS 3921 w13x for a period of 5 min for each sample with gravimetric moisture determinations taken at 1-min intervals. Secondly, the same samples were dried and repeat-tested, as described above, but instead using the IRS ‘wick’ test apparatus. Hall w14x previously observed that no changes occurred in the pore structure of fired clay bricks during repeat-tests of this nature. The IRS ‘wick’ test values were in close agreement with those obtained using the BS 3921 IRS test. Fig. 5 illustrates the comparison between BS 3921 IRS and
IRS ‘wick’ test data obtained during testing. Note that the mass of sorbed water in the ‘wick’ test samples is around half that of the BS 3921 IRS test samples due to the inflow-surface area also being approximately half that of the latter. A good correlation is indicated by a similar IRS value, calculated as the mass of sorbed water per unit of surface area over time. It was observed that for very porous samples a greater disparity occurred between values obtained using the IRS test and the IRS ‘wick’ test. However, rammed earth is particularly dense with a low porosity and so the IRS ‘wick’ test is considered a suitable means with which to test this material. The variation that occurs between results for individual test specimens is significant and is an inherent problem with both the BS 3921 IRS test and the IRS ‘wick’ test. The average level of variation between rammed earth samples was observed to be between approximately 20 and 40%. These parameters are well within the typical variations observed in our laboratory for fired clay bricks after several thousand cycles of BS 3921 IRS testing. This emphasises the myopic nature of testing masonry building materials at this scale when, for example, the behaviour of a construction element (e.g. a wall) cannot be defined by the performance of an individual brick. However, testing a randomly selected batch of samples can be used to give a good indication of a porous masonry materials performance in terms of moisture ingress due to capillary suction through a set of average values. 5. Rammed earth testing The authors have recently proposed recommendations for the production of rammed earth cube samples using synthetically blended soils whose particle-size distribution can accurately be specified and controlled w1x. The
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
273
Fig. 5. Comparison between BS 3921 IRS test and IRS ‘wick’ test results.
same ten soil recipes, specified previously, have been used for the experimental work detailed in this article, as can be seen from the particle-size distribution chart illustrated in Fig. 6. Further details on the physical and mechanical properties of the rammed earth cube samples have been previously illustrated, by the authors, in depth w1x. This method of rammed earth sample production
was found highly repeatable and consistent in terms of density, dimensional stability and compressive strength. The rammed earth samples were tested in accordance with the previously defined IRS ‘wick’ test, which is a peripheral to the BS 3921 IRS test. Six samples from each of the ten soil types identified above were used as the test specimens. Each sample was tested for a period
Fig. 6. BS 1377 Particle-size distribution chart with rammed earth soil data.
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
274
Table 1 The average amount of sorbed water during a 5-min IRS ‘wick’ test Masonry type
Dry massavg (g)
Average wm (g)
Average wm (%)
London brick – Fletton London brick – Dapple light Engineering brick C30 concrete
851.0 923.2 1316.6 2165.8
10.95 7.12 12.47 6.72
1.29 0.77 0.95 0.31
Rammed Rammed Rammed Rammed Rammed Rammed Rammed Rammed Rammed Rammed
2136.9 2132.0 2068.8 2030.6 2180.5 2101.3 2120.0 2058.4 2067.3 2089.0
3.17 3.37 8.45 15.15 3.42 4.22 4.15 5.95 5.92 3.12
0.15 0.16 0.41 0.74 0.16 0.20 0.20 0.29 0.28 0.15
earth earth earth earth earth earth earth earth earth earth
– – – – – – – – – –
532 622 712 802 433 523 613 703 424 514
of 5-min and the mass of sorbed water (mw) recorded at 1-min intervals. After testing the samples were then allowed to dry before the IRS ‘wick’ test regime (described above) was repeated using the same samples and the same inflow-surfaces. This time, however, only three from each of the six samples (in each set) were selected for the re-testing. The samples selected from a particular soil type all had a similar IRS value, thereby eliminating the upper and lower values from the repeattest results. Weight determinations were made at 1-min intervals but the total test duration was extended to give values for 1, 2, 3, 4, 5, 10, 20, 30, 40, 50 and 60-min elapsed time intervals. 6. Results and discussion Rammed earth generally has a very low initial rate of suction compared to conventional masonry materials such as concrete and fired clay bricks, and absorbs much smaller amounts of water over a given time span. This is perhaps due to its relatively high density and resultant lower bulk porosity, as defined earlier. This has been clearly illustrated in Table 1 where the average mass of sorbed water, during a 5-min IRS ‘wick’ test, for each material tested has been expressed as a percentage of the sample’s dry mass. The initial rate of suction over a period of 5 min has been illustrated in Fig. 7 to show the comparison between rammed earth and conventional masonry materials. Fletton bricks and Engineering bricks, for example, have a higher suction rate than every rammed earth sample apart from 802. C30 concrete is just above 703, which is the third poorest rammed earth sample. The fired clay brick IRS curves are a different gradient to rammed earth, but concrete is almost the same indicating similarities between their internal pore structure. The large degree of variation that occurred between individual test specimens was attributed to natural variations in pore structure both
within the cube and, more importantly, at the inflow test face. The surface finish of the inflow test face appears to be critical in determining the level of moisture ingress in rammed earth. It is controlled by the particle-size distribution of the soil and by the random location of different sized particles towards the facade. ¸ Larger particles in this region appear to cause micro-cracking and internal fissures that allow large volumes of water to be imbibed. In addition, moisture ‘tracking’ may occur where the migrating moisture follows the contour of larger (e.g. gravel) particles, within the matrix, due to their surface tension. This factor may be magnified on small-scale laboratory tests such as this where, for example, a piece of 14 mm gravel represents over 1y 8th the width of an 80 mm inflow surface test area. In addition, moisture ingress in rammed earth facades ¸ was observed to be much greater at the compaction planes (i.e. the zones between layers of compacted soil) where the inherent degree of compaction is slightly less. Discretionary operator control during small-scale laboratory testing introduces factors that would not normally be present in the testing of a full-sized wall. Sample faces that were free from defects, for example, were selected for testing to reduce the variability of results and random effects. This level of control is acceptable for small-scale laboratory testing, as the number of variables should be minimised wherever possible. The use of full-sized test walls would lessen these considerations and, in one sense, provide a more realistic representation of the materials performance insitu. However, accuracy and quality control tend to diminish as the scale of the test increases. Fig. 8 shows the comparison between the ten rammed earth soil types tested over a 60-min duration. The IRS curves exist in three distinct bands: high (802), medium (712, 703, 424) and low (523, 613, 433, 622, 532, 514). Beyond the 10-min point, the IRS of all samples
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
275
Fig. 7. IRS ‘wick’ test results (5-min regime) for rammed earth vs. conventional masonry.
begins to decrease significantly and by 50–60 min, all samples are at a very similar IRS value. Notably, it has been observed previously that after approximately 3 h of testing mw is proportional to t giving experimental confirmation of a steady state in three-dimensional water absorption scenarios such as this. The ‘wick’ test is three-dimensional because Hall w14x previously observed that, in a modified IRS test, a spherical inflow source in the centre of sample face allows for lateral spreading of the sorbed water within the material. It is possibly more realistic than the one-dimensional BS 3921 IRS test for experimentally determining the ingress of moisture, due to capillary suction, and comparing the results obtained to those for full-sized test walls. Furthermore, the geometry and dimensions of inflow-surface area on the IRS ‘wick’ test can be transposed to other threedimensional moisture ingress test methodologies, such as the Initial Surface Absorption (ISA) test, to allow for comparison between the obtained data. Differences were observed, in the IRS of a given sample, between being tested in the 5-min regime, and
being re-tested in the 60-min regime. Re-testing rammed earth appears to result in lowering its IRS value. The average variation between results, due to repeat testing, indicates an approximate decrease in IRS value from as little as 5% up to as much as 44%. The quantity of reduction in IRS appears to coincide with samples that absorb more water due to capillary suction, as can be clearly seen from Fig. 9. This indicates that the ingress of water may have changed the properties of the material, such as its internal pore structure, especially since the surface finish had visibly become altered following the first regime of testing. Theoretically the yt law (detailed above) is only applicable for the one-dimensional case of the BS 3921 IRS test where the inflow of sorbed water is normal to the inflow surface w14x. Hall states that for the circular source, with three-dimensional lateral internal spreading (i.e. IRS ‘wick’ test), cumulative absorption should increase more rapidly than yt for common clay bricks w14x. However, the results obtained for rammed earth show that when the mass of sorbed water wmw(g)x is
276
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 8. IRS ‘wick’ repeat-test results (60-min regime) for rammed earth.
w
z
plotted against the root of time xyytŽmin.|~ the relationship is clearly linear as can be seen from Fig. 10. This may be because the total mass of sorbed water is considerably less in rammed earth, and the internal pore structure is quite different. The linear value allows for interpolation and future forecast of the amount of water taken up by a given mass of rammed earth after a given elapsed time. It also shows the performance of each soil type more clearly allowing direct comparisons to be made more easily. Notably, the same three bands appear: high, medium and low, as described earlier. The bulk porosity of the samples was calculated by first determining the particle density of each soil type in accordance with BS 1377 w18x. The calculation of void space (bulk porosity) within a rammed earth sample can thus be calculated when its dry mass and volumetric proportions are known. Fig. 11 illustrates the positive relationship that exists between bulk porosity and the IRS value taken after 1 min of testing. The bulk porosity (total void space) of rammed earth is therefore likely to be a similar value to the apparent porosity, i.e. the amount of void space that is permeable and not enclosed. This indicates that much of the pore space within rammed earth is possibly interlinked and potentially permeable to moisture. Linear increasesydecreases in the compressive strength of rammed earth, however,
were found to give no obvious pattern or relationship with the IRS value. The total specific surface area (SSA) value for each soil type was calculated using the retained mass values obtained during sieve analysis. Assuming each individual soil particle to be spherical, its surface area can be simply calculated using 4pr2. The mass retained for each particle diameter is known; therefore the SSA can be calculated and expressed in millimetre square per gram. The total SSA for a given soil type is simply the summation of the SSA values obtained for each sieve size. The author hypothesised that the binder is not as effectively distributed amongst the aggregate particles when the specific surface area of the latter is increased beyond a certain value. This may result in the enhanced ability of the aggregate to adsorb moisture, due to surface tension, resulting in increased capillary potential within the matrix. It may also lead to areas of pore space that are devoid of binder leaving few obstructions to impede the flow of imbibed water. This decrease in effective binder distribution within the matrix of the material may lead to a more open-structured pore network capable of allowing a greater rate of moisture ingress and subsequent migration. In support of this theory, Figs. 12 and 13 show an increase in average IRS value coinciding with a relative increase in the mass of particles with diameters between 63 um and 3.35 mm,
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
277
Fig. 9. Variation of IRS in rammed earth samples due to single repeat-testing.
but with a sharp decrease in the mass of particles above 3.35 mm. This may indicate that a rise in the number of particles )3.35 mm can give a sufficient decrease in total SSA such that the effectiveness of binder distribution returns to a level where capillary potential is significantly reduced. It can be observed that when the ratio of )3.35 to -3.35 mm particles (defined here as the ‘3.35 ratio’) is greater than 5, the amount of water imbibed due to capillary suction increases to measurably significant levels. This observation has been clearly illustrated in Fig. 14. 7. Conclusions The novel IRS ‘wick’ test has been devised as a peripheral to the established BS 3921 IRS test, and is
suitable for use with unstabilised earth materials that slake on contact with water. It produces consistent results that, for dense samples such as rammed earth, are in good agreement with the original BS 3921 IRS test. Rammed earth generally outperforms conventional modern masonry materials significantly in terms of the rate and quantity of moisture ingress due to capillary suction. However, the particle-size distribution of the soil is critical in determining the rate at which moisture may ingress due to capillary suction. The ingress of moisture into rammed earth appears to alter its properties such that in a repeat test the IRS is reduced. The amount by which the IRS value is reduced in a repeat test appears to be proportionally indexed to the original IRS value prior to repeat testing.
Fig. 10. Relationship between mass of sorbed water and square root of elapsed time.
278
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 11. Relationship between the bulk porosity of rammed earth and its IRS value.
Moisture ingress in rammed earth, due to capillary suction, increases linearly per unit inflow surface area against the square root of elapsed time Žyt. . This allows predictions to be made on the rate and amount of moisture ingress at a given future point in time. The results obtained are in good agreement with current nonsaturated flow theory further supporting the validity of the proposed IRS ‘wick’ test. In a suitable soil, the ratio between the total specific surface area (SSA) of the aggregate fraction and the mass of the binder fraction appears to be positively
linked with the rate of capillary suction in rammed earth. When the mass of the binder fraction in a suitable soil is less than 10% of the total soil mass, it would appear that the rate of moisture ingress, due to capillary suction, in rammed earth is significantly increased. It is hypothesised, therefore, that through granular stabilisation (i.e. artificially modifying the particle-size distribution through additionysubtraction of material) the rate of capillary moisture ingress in rammed earth can be controlled. A simple initial suggestion may be to ensure a minimum binder quantity of approximately 10% by
Fig. 12. Particle-size distribution and IRS values for rammed earth soils with two parts binder.
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
279
Fig. 13. Particle-size distribution and IRS values for rammed earth soils with three parts binder.
Fig. 14. Relationship between IRS value and ‘3.35 ratio’ for rammed earth soils containing two and three parts binder.
mass. A ‘3.35 ratio’ of 5 has been proposed as a potential target value by which to attenuate the required approach to granular stabilisation. The 802 soil (worst), for example, can have its ‘3.35 ratio’ lowered from above 20.0 to below 5.0 by adding 12% by mass of 10 mm pea gravel. This should have the effect of lowering the rate of moisture ingress due to capillary suction, or even achieving an optimum value. This observation has significant practical applications but requires testing in the field in order to prove or disprove the hypothesis. Acknowledgments The author wishes to acknowledge the help of the following parties: Stephen Dobson (RAMTEC Pty Ltd)
for pictures, helpful advice and discussion Stephen Hetherington (Sheffield Hallam University) for technical support Paul Scholey (Rotherham Sand & Gravel Co Ltd.) for generous donation of earth materials. References w1x Hall M, Djerbib Y. Rammed earth sample production: context, recommendations and consistency. Constr Build Mater 2004;118(4):281 –286. w2x Dobson S. (RAMTEC Pty Ltd.) Pers. comm. w3x Oliver A. Dampness in buildings. Second ed. Oxford: Blackwells, 1997. w4x Building research establishment (BRE). The selection of natural building stone. London: HMSO, 1983. p. 1 –8 (Digest 269).
280
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
w5x Laycock EA. Frost degradation and weathering of the magnesian limestone building stone of the Yorkshire province. Ph.D. Thesis, University of Sheffield, 1997. w6x Hall C. Water movement in porous building materials – I: unsaturated flow theory and its applications. Build Environ 1977;12:117 –25. w7x Killip IR, Cheetham DW. The prevention of rain penetration through external walls and joints by means of pressure equalisation. Build Environ 1984;19(2):81 –91. w8x Camuffo D. Physical weathering of stones. Elsevier Science, 1995. p. 1 –14. w9x Vos BH, Tammes E. Flow of water in the liquid phase. Report no. B 1-68-38, Inst. TNO for building materials and building structures, Delft, Holland, 1968. w10x Kieslinger A. Feuchtigkeitsschaden ¨ an bauwerken. Zemet Beton 1957;9:1 –7. w11x Mamillan. Connaissances actuelles des problemes de remontees ´ d’eau par capillarite´ dans les mures. Rapporti della soprintendenza per I beni artistici e storica per le province di bologna, Ferrara, forli e ravenna. The conservation of stone II pre-prints of the contributions to the international symposium, Bologna,
w12x
w13x w14x
w15x
w16x w17x
w18x
27–30 October 1981: part A deterioration, Rossi-manaresi R. (ed.), Bologna, pp. 59–72. Vos BH, Tammes E. Moisture and moisture transfer in porous materials. Report no. BI-69-96, Nov 1969. Organisation for industrial research TNO, Institute TNO for building materials and building structures, Delft, Holland. British Standards Institute (BSI). BS 3921: 1985 (partially withdrawn) specification for clay bricks. BSI: London, 1985. Hall C. Water movement in porous building materials – IV: the initial surface absorption and the sorptivity. Build Environ 1981;16(3):117 –25. Hall C, Kam-Ming Tse T. Water movement in porous building materials – VII: the sorptivity of mortars. Build Environ 1986;21(2):113 –8. Minke G. Earth construction handbook: the building material earth in modern architecture. UK: WIT Press, 2000. De Freitas VP, Abrantes V, Crausse P. Moisture migration in building walls – analysis of the interface phenomena. Build Environ 1996;31(2):99 –108. British Standards Institute (BSI). BS 1377-2: 1990 soils for civil engineering purposes – part 2: classification. BSI: London, 1990.
Moisture ingress in rammed earth: Part 1—the effect of soil particle-size distribution on the rate of capillary suction Matthew Hall*, Youcef Djerbib Centre for the Built Environment, Sheffield Hallam University, Unit 9 Science Park, Howard Street, Sheffield S1 1WB, UK Received 19 November 2003; received in revised form 24 November 2003; accepted 24 November 2003
Abstract The novel initial rate of suction (IRS) ‘wick’ test has been presented, and is suitable for determining the rate of capillary moisture ingress in unstabilised rammed earth that slakes on contact with water. Experimental testing was performed using the ‘wick’ test to investigate the effect of soil particle-size distribution on moisture ingress in rammed earth. Rammed earth generally absorbs much less water due to capillary suction, and at a slower rate, than conventional masonry building materials such as bricks and concrete. Moisture ingress in rammed earth, due to capillary suction, increases linearly per unit inflow surface area against the square root of elapsed time yt . The particle-size distribution of the soil is critical in determining the rate at which moisture may ingress. In a suitable soil, the ratio between the total specific surface area (SSA) of the aggregate fraction and the mass of the binder fraction appears to be positively linked with the rate of capillary suction in rammed earth. Experimental data have been included. 䊚 2003 Elsevier Ltd. All rights reserved.
Ž .
Keywords: Rammed earth; Moisture ingress; Capillary suction
1. Introduction Rammed earth can be defined as ‘«an ancient earth building technique that involves dynamically compacting moist sub-soil between removable shuttering to create an in-situ monolithic compressed earth wall that is both strong and durable’ w1x. Figs. 1 and 2 show a rammed earth dwelling constructed by RAMTEC Pty Ltd in Western Australia, illustrating the current levels of sophistication in design and application of this natural, sustainable material. Some current public perceptions of rammed earth as being a material with low climatic durability are misguided but not entirely unfounded. Rammed earth construction is popular and performs well in warm, dry climates. However, research is essential in determining the application of rammed earth in temperate damp climates if it is to succeed in counties such as the UK. Cavity wall masonry construction, which is by far the most common form in Britain, has largely eliminated the problems of penetrating damp*Corresponding author. Tel.: q44-0-114-225-3200; fax: q44-0114-225-3206. E-mail address: [email protected] (M. Hall).
ness that often plagued the solid walls of older buildings. As rammed earth walls are monolithic, the capillary movement of moisture within them is a particular problem, especially since there are no chemically dosed layers of mortar to suppress it w2x. A variety of chemical additives and admixtures that are available commercially could be used to alleviate the problem but current data relating to their effectiveness is not forthcoming and so poses serious problems for earth building contractors. These problems may be exacerbated depending upon the composition of the soil mix used to form the rammed earth walls; the first logical step is to understand the material before investigating how best to treat it. We have tried many things before regarding admixtures for damp ingress-all too expensive or unreliable«Does adding clay make it better or worse? Does removing clay (e.g. by washing) make it better or worse? Does lessening clay% (e.g. by adding washed sand) make it better or worse? These are real questions w2x.
2. Moisture ingress by capillary suction Excessive moisture content in building elements is referred to as damp w3x. Dampness can enter into and
0950-0618/04/$ - see front matter 䊚 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2003.11.002
270
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 1. Modern rammed earth house in Western Australia (Courtesy RAMTEC Pty Ltd 䉷2001).
move within the fabric of porous construction materials by a variety of different processes. Porous materials are often permeable by air and moisture because they contain a network of open channels. According to the Building Research Establishment w4x, there are two classifications for these channels dependent upon their diameter. A channel with a nominal diameter of 5 mm or above is classed as a pore whereas a channel with a nominal diameter of less than this value is classed as a capillary or a micropore. It should be noted that pores and capillaries are not cylindrical but in fact often exhibit a high level of tortuosity w5x. They may also be interlinked; the extent of which largely determines the permeability of a porous material. Hall w6x observed that if, hypothetically, the solid material of a porous building material is assumed to be inert (i.e. no mass transfer occurs), and that specific boundary effects and dissolved atmospheric gases in the water are ignored a considerable simplification ensues. The role of the solid porous material is then reduced to acting as a pore network that ‘partially constrains the boundary surfaces of the water phases’ to a particular geometry. The system then contains a single component—water—that may exist in any combination of solid, liquid or gaseous form. The specific moisture content w of this idealised porous solid can thus be
Fig. 2. Inside a modern rammed earth home (Courtesy RAMTEC Pty Ltd 䉷2001).
determined, where mw denotes the mass of water and ms the mass of the dry solid: ws
mw . ms
(1)
In practice, the value of w is determined gravimetrically as the difference between the total weights of the sample when it is oven dried to a constant weight at 105 8C, and when it is wet. The conventional description of the reference moisture content of a porous material allows us to clearly distinguish between the hygroscopic moisture content and the capillary moisture content, as can be seen from Fig. 3. The theoretical point of total saturation (Wmax) is rarely achieved, and neither is a completely water-free state (0). In the nominal dry state a degree of hygroscopic water remains, defined by the point Wh, whereas above the nominal point of saturation (Wsat) techniques such as vacuum saturation are required in order to achieve higher levels of moisture content. Movement of moisture in walls is controlled by the masonry and by discontinuities (e.g. fracturing) within the material w5x. In terms of the mechanisms of moisture movement we are dealing with ‘«a single fundamental process, the movement of water through a permeable material whose water content is non-uniform and gen-
Fig. 3. Reference moisture contents, adapted from de Freitas et al. (1996).
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
erally less than saturation’ w6x. To illustrate the process of moisture ingress by capillary suction, Hall w6x used the classic example of a dry brick that has been placed in a shallow tray containing clean water (i.e. the initial rate of suction test). From the moment this is done water is absorbed into the brick by capillary action, and then attempts to distribute itself throughout the pore network of the brick itself. The imbibed water partially displaces the air that previously occupied the pores, whilst at the surfaces that are not actually in contact with water evaporation has begun to occur. At this point, equilibrium can be achieved between evaporative loss and water absorption. However, evaporation entails cooling at the boundary surfaces, which results in a heat flow process being generated inside the brick. The resultant heat gradient modifies the rate of flow of moisture inside the brick. Soluble salts are often dissolved and redeposited at the surface where evaporation is occurring, and the resultant crystalline deposits referred to as efflorescence. As a material approaches saturation, the capillary suction approaches zero w7x. Capillary suction in porous materials, therefore, occurs by the progressive displacement of water so that pockets of air entrapped in the pore and capillary system cannot stop it w8x. The only exception to this occurs where the capillary pressure is simply counteracted by the pressure of entrapped air in a duct that is sealed and has no outlet. Vos and Tammes w9x performed various experiments on the capillary movement of water and concluded that water moving through a porous material by capillarity could travel twice as far in a horizontal direction than in a vertical direction, owing to the effects of gravity. Kieslinger w10x noted that a good correlation between field data and the equation for determining the height of capillary rise had been observed: hs
2s Rd
(2)
where h: water height achieved (cm); s: (constant) 0.074 gycm3; d: density of water (approx. 1 gycm3); R: radius of capillary (cm). Capillary rise is therefore dependent on the diameter of the tube and the liquid within the system. Mamillan w11x gives the following example where, in theory, a 1 mm pore could enable water to rise by 15 mm; a 0.1mm pore enables a capillary rise of 150 mm; and a 0.01 mm pore enables a rise of 1500 mm. The durability of many types of natural building stones has been related to pore size w5x, such that those with high microporosity are often less durable than those with a lower content of micropores. According to Vos and Tammes w12x, if a temperature gradient exists along a capillary then water flow will occur in the direction of lower temperature. This is
271
because surface tension of the capillary wall material is a ‘monotonously decreasing function of temperature’. Conversely, water held within a capillary of uniform temperature but with a gradient in diameter will flow towards the narrower diameter. Consequently, if two capillaries communicate with one another moisture transfer will occur between the material with larger pore diameters to that with the smaller pore diameters until the gradient of unequal surface tension has been resolved through equilibrium w12x. 3. Initial rate of suction (IRS) test The British Standard BS 3921 w13x test for determining capillary suction of water into masonry materials is known as the initial rate of suction (IRS) test. The test method is essentially a gravimetric determination of sorbed water in a partially immersed porous building material (due to capillary suction) over time, the value for which is expressed in kilograms per meter square minute. Hall w14x previously observed the cumulative absorbed mass of water (mw) per unit area of the inflow surface, increases linearly against the square root of elapsed time Žyt.. The sample is to be kept ‘dry’ prior to testing. The current IRS test is, however, of little use for materials that slake (e.g. earth materials) because the determination of imbibed water mass requires the mass of the sample to be kept constant. Furthermore, the test requires the wetted sample surface to be wiped with a damp cloth to remove excess water. This presents a large degree of operator error and a potential drying effect depending upon the type, and moisture content, of the cloth. The testing procedure can be enhanced by maintaining a constant water temperature of 20 8C ("18) in order to ensure constant viscosity. Hall and Kam-Ming Tse w15x recommend that the weighing operation be completed as quickly as possible, ideally within 30 s. Minke w16x has discussed the German standard version of the IRS test, DIN 52 617, within the context of earth building materials. The main differences involve encasing the sample cube sides in fibre-reinforced polyester resin, gluing filter paper to the test face and then placing it on a submerged polyurethane foam base. However, this level of sample intervention is perhaps excessive and deviates significantly from the ‘natural’ conditions of the BS 3921 IRS test. The impermeable resin coating constricts the natural expansion of the sample and restricts the displacement of air from the pore network caused by the imbibed water. Furthermore, equilibrium between evaporative loss at the sample faces and water absorption cannot occur. 4. The IRS ‘wick’ test The IRS ‘wick’ test is a novel adaptation of the current BS 3921 IRS test apparatus (see Fig. 4) devised
272
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 4. BS 3921 IRS test apparatus with IRS ‘wick’ test peripherals.
by the author (Hall). A 30-mm high Oasis娃 ‘wick’, which is 80 mm in diameter, acts as the point of contact for the sample and represents free water being sorbed from another saturated porous medium that offers negligible capillary resistance. However, according to the studies on interface phenomena performed by De Freitas, Abrantes and Crausse w17x, theoretically there should be a small degree of hydric resistance that conditions the maximum flow of moisture transmitted from the wick to the sample face being tested. The contact area is constant and there are no edge effects or meniscus errors on inflow-surface area calculations, as was previously the case in the BS 3921 IRS test. Materials that slake in contact with water are stable throughout the test and negligible mass loss occurs as the inflow surface is retained by the self-weight of the sample acting on the solid surface of the wick. The wick itself can simply be taken out, washed and reused to maintain accuracy and repeatability. Testing was performed on C30 concrete 100=100mm cubes and 3 types of brick; London Brick ‘Fletton’ (high porosity), London Brick ‘Dapple Light’ (medium porosity) and Engineering Brick (low porosity). Six representative samples of each material type were selected. The bricks were cut on a diamond bit saw to give a 100=100-mm inflow-surface area the same as the cube samples. Firstly, the IRS test was performed in accordance with BS 3921 w13x for a period of 5 min for each sample with gravimetric moisture determinations taken at 1-min intervals. Secondly, the same samples were dried and repeat-tested, as described above, but instead using the IRS ‘wick’ test apparatus. Hall w14x previously observed that no changes occurred in the pore structure of fired clay bricks during repeat-tests of this nature. The IRS ‘wick’ test values were in close agreement with those obtained using the BS 3921 IRS test. Fig. 5 illustrates the comparison between BS 3921 IRS and
IRS ‘wick’ test data obtained during testing. Note that the mass of sorbed water in the ‘wick’ test samples is around half that of the BS 3921 IRS test samples due to the inflow-surface area also being approximately half that of the latter. A good correlation is indicated by a similar IRS value, calculated as the mass of sorbed water per unit of surface area over time. It was observed that for very porous samples a greater disparity occurred between values obtained using the IRS test and the IRS ‘wick’ test. However, rammed earth is particularly dense with a low porosity and so the IRS ‘wick’ test is considered a suitable means with which to test this material. The variation that occurs between results for individual test specimens is significant and is an inherent problem with both the BS 3921 IRS test and the IRS ‘wick’ test. The average level of variation between rammed earth samples was observed to be between approximately 20 and 40%. These parameters are well within the typical variations observed in our laboratory for fired clay bricks after several thousand cycles of BS 3921 IRS testing. This emphasises the myopic nature of testing masonry building materials at this scale when, for example, the behaviour of a construction element (e.g. a wall) cannot be defined by the performance of an individual brick. However, testing a randomly selected batch of samples can be used to give a good indication of a porous masonry materials performance in terms of moisture ingress due to capillary suction through a set of average values. 5. Rammed earth testing The authors have recently proposed recommendations for the production of rammed earth cube samples using synthetically blended soils whose particle-size distribution can accurately be specified and controlled w1x. The
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
273
Fig. 5. Comparison between BS 3921 IRS test and IRS ‘wick’ test results.
same ten soil recipes, specified previously, have been used for the experimental work detailed in this article, as can be seen from the particle-size distribution chart illustrated in Fig. 6. Further details on the physical and mechanical properties of the rammed earth cube samples have been previously illustrated, by the authors, in depth w1x. This method of rammed earth sample production
was found highly repeatable and consistent in terms of density, dimensional stability and compressive strength. The rammed earth samples were tested in accordance with the previously defined IRS ‘wick’ test, which is a peripheral to the BS 3921 IRS test. Six samples from each of the ten soil types identified above were used as the test specimens. Each sample was tested for a period
Fig. 6. BS 1377 Particle-size distribution chart with rammed earth soil data.
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
274
Table 1 The average amount of sorbed water during a 5-min IRS ‘wick’ test Masonry type
Dry massavg (g)
Average wm (g)
Average wm (%)
London brick – Fletton London brick – Dapple light Engineering brick C30 concrete
851.0 923.2 1316.6 2165.8
10.95 7.12 12.47 6.72
1.29 0.77 0.95 0.31
Rammed Rammed Rammed Rammed Rammed Rammed Rammed Rammed Rammed Rammed
2136.9 2132.0 2068.8 2030.6 2180.5 2101.3 2120.0 2058.4 2067.3 2089.0
3.17 3.37 8.45 15.15 3.42 4.22 4.15 5.95 5.92 3.12
0.15 0.16 0.41 0.74 0.16 0.20 0.20 0.29 0.28 0.15
earth earth earth earth earth earth earth earth earth earth
– – – – – – – – – –
532 622 712 802 433 523 613 703 424 514
of 5-min and the mass of sorbed water (mw) recorded at 1-min intervals. After testing the samples were then allowed to dry before the IRS ‘wick’ test regime (described above) was repeated using the same samples and the same inflow-surfaces. This time, however, only three from each of the six samples (in each set) were selected for the re-testing. The samples selected from a particular soil type all had a similar IRS value, thereby eliminating the upper and lower values from the repeattest results. Weight determinations were made at 1-min intervals but the total test duration was extended to give values for 1, 2, 3, 4, 5, 10, 20, 30, 40, 50 and 60-min elapsed time intervals. 6. Results and discussion Rammed earth generally has a very low initial rate of suction compared to conventional masonry materials such as concrete and fired clay bricks, and absorbs much smaller amounts of water over a given time span. This is perhaps due to its relatively high density and resultant lower bulk porosity, as defined earlier. This has been clearly illustrated in Table 1 where the average mass of sorbed water, during a 5-min IRS ‘wick’ test, for each material tested has been expressed as a percentage of the sample’s dry mass. The initial rate of suction over a period of 5 min has been illustrated in Fig. 7 to show the comparison between rammed earth and conventional masonry materials. Fletton bricks and Engineering bricks, for example, have a higher suction rate than every rammed earth sample apart from 802. C30 concrete is just above 703, which is the third poorest rammed earth sample. The fired clay brick IRS curves are a different gradient to rammed earth, but concrete is almost the same indicating similarities between their internal pore structure. The large degree of variation that occurred between individual test specimens was attributed to natural variations in pore structure both
within the cube and, more importantly, at the inflow test face. The surface finish of the inflow test face appears to be critical in determining the level of moisture ingress in rammed earth. It is controlled by the particle-size distribution of the soil and by the random location of different sized particles towards the facade. ¸ Larger particles in this region appear to cause micro-cracking and internal fissures that allow large volumes of water to be imbibed. In addition, moisture ‘tracking’ may occur where the migrating moisture follows the contour of larger (e.g. gravel) particles, within the matrix, due to their surface tension. This factor may be magnified on small-scale laboratory tests such as this where, for example, a piece of 14 mm gravel represents over 1y 8th the width of an 80 mm inflow surface test area. In addition, moisture ingress in rammed earth facades ¸ was observed to be much greater at the compaction planes (i.e. the zones between layers of compacted soil) where the inherent degree of compaction is slightly less. Discretionary operator control during small-scale laboratory testing introduces factors that would not normally be present in the testing of a full-sized wall. Sample faces that were free from defects, for example, were selected for testing to reduce the variability of results and random effects. This level of control is acceptable for small-scale laboratory testing, as the number of variables should be minimised wherever possible. The use of full-sized test walls would lessen these considerations and, in one sense, provide a more realistic representation of the materials performance insitu. However, accuracy and quality control tend to diminish as the scale of the test increases. Fig. 8 shows the comparison between the ten rammed earth soil types tested over a 60-min duration. The IRS curves exist in three distinct bands: high (802), medium (712, 703, 424) and low (523, 613, 433, 622, 532, 514). Beyond the 10-min point, the IRS of all samples
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
275
Fig. 7. IRS ‘wick’ test results (5-min regime) for rammed earth vs. conventional masonry.
begins to decrease significantly and by 50–60 min, all samples are at a very similar IRS value. Notably, it has been observed previously that after approximately 3 h of testing mw is proportional to t giving experimental confirmation of a steady state in three-dimensional water absorption scenarios such as this. The ‘wick’ test is three-dimensional because Hall w14x previously observed that, in a modified IRS test, a spherical inflow source in the centre of sample face allows for lateral spreading of the sorbed water within the material. It is possibly more realistic than the one-dimensional BS 3921 IRS test for experimentally determining the ingress of moisture, due to capillary suction, and comparing the results obtained to those for full-sized test walls. Furthermore, the geometry and dimensions of inflow-surface area on the IRS ‘wick’ test can be transposed to other threedimensional moisture ingress test methodologies, such as the Initial Surface Absorption (ISA) test, to allow for comparison between the obtained data. Differences were observed, in the IRS of a given sample, between being tested in the 5-min regime, and
being re-tested in the 60-min regime. Re-testing rammed earth appears to result in lowering its IRS value. The average variation between results, due to repeat testing, indicates an approximate decrease in IRS value from as little as 5% up to as much as 44%. The quantity of reduction in IRS appears to coincide with samples that absorb more water due to capillary suction, as can be clearly seen from Fig. 9. This indicates that the ingress of water may have changed the properties of the material, such as its internal pore structure, especially since the surface finish had visibly become altered following the first regime of testing. Theoretically the yt law (detailed above) is only applicable for the one-dimensional case of the BS 3921 IRS test where the inflow of sorbed water is normal to the inflow surface w14x. Hall states that for the circular source, with three-dimensional lateral internal spreading (i.e. IRS ‘wick’ test), cumulative absorption should increase more rapidly than yt for common clay bricks w14x. However, the results obtained for rammed earth show that when the mass of sorbed water wmw(g)x is
276
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 8. IRS ‘wick’ repeat-test results (60-min regime) for rammed earth.
w
z
plotted against the root of time xyytŽmin.|~ the relationship is clearly linear as can be seen from Fig. 10. This may be because the total mass of sorbed water is considerably less in rammed earth, and the internal pore structure is quite different. The linear value allows for interpolation and future forecast of the amount of water taken up by a given mass of rammed earth after a given elapsed time. It also shows the performance of each soil type more clearly allowing direct comparisons to be made more easily. Notably, the same three bands appear: high, medium and low, as described earlier. The bulk porosity of the samples was calculated by first determining the particle density of each soil type in accordance with BS 1377 w18x. The calculation of void space (bulk porosity) within a rammed earth sample can thus be calculated when its dry mass and volumetric proportions are known. Fig. 11 illustrates the positive relationship that exists between bulk porosity and the IRS value taken after 1 min of testing. The bulk porosity (total void space) of rammed earth is therefore likely to be a similar value to the apparent porosity, i.e. the amount of void space that is permeable and not enclosed. This indicates that much of the pore space within rammed earth is possibly interlinked and potentially permeable to moisture. Linear increasesydecreases in the compressive strength of rammed earth, however,
were found to give no obvious pattern or relationship with the IRS value. The total specific surface area (SSA) value for each soil type was calculated using the retained mass values obtained during sieve analysis. Assuming each individual soil particle to be spherical, its surface area can be simply calculated using 4pr2. The mass retained for each particle diameter is known; therefore the SSA can be calculated and expressed in millimetre square per gram. The total SSA for a given soil type is simply the summation of the SSA values obtained for each sieve size. The author hypothesised that the binder is not as effectively distributed amongst the aggregate particles when the specific surface area of the latter is increased beyond a certain value. This may result in the enhanced ability of the aggregate to adsorb moisture, due to surface tension, resulting in increased capillary potential within the matrix. It may also lead to areas of pore space that are devoid of binder leaving few obstructions to impede the flow of imbibed water. This decrease in effective binder distribution within the matrix of the material may lead to a more open-structured pore network capable of allowing a greater rate of moisture ingress and subsequent migration. In support of this theory, Figs. 12 and 13 show an increase in average IRS value coinciding with a relative increase in the mass of particles with diameters between 63 um and 3.35 mm,
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
277
Fig. 9. Variation of IRS in rammed earth samples due to single repeat-testing.
but with a sharp decrease in the mass of particles above 3.35 mm. This may indicate that a rise in the number of particles )3.35 mm can give a sufficient decrease in total SSA such that the effectiveness of binder distribution returns to a level where capillary potential is significantly reduced. It can be observed that when the ratio of )3.35 to -3.35 mm particles (defined here as the ‘3.35 ratio’) is greater than 5, the amount of water imbibed due to capillary suction increases to measurably significant levels. This observation has been clearly illustrated in Fig. 14. 7. Conclusions The novel IRS ‘wick’ test has been devised as a peripheral to the established BS 3921 IRS test, and is
suitable for use with unstabilised earth materials that slake on contact with water. It produces consistent results that, for dense samples such as rammed earth, are in good agreement with the original BS 3921 IRS test. Rammed earth generally outperforms conventional modern masonry materials significantly in terms of the rate and quantity of moisture ingress due to capillary suction. However, the particle-size distribution of the soil is critical in determining the rate at which moisture may ingress due to capillary suction. The ingress of moisture into rammed earth appears to alter its properties such that in a repeat test the IRS is reduced. The amount by which the IRS value is reduced in a repeat test appears to be proportionally indexed to the original IRS value prior to repeat testing.
Fig. 10. Relationship between mass of sorbed water and square root of elapsed time.
278
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
Fig. 11. Relationship between the bulk porosity of rammed earth and its IRS value.
Moisture ingress in rammed earth, due to capillary suction, increases linearly per unit inflow surface area against the square root of elapsed time Žyt. . This allows predictions to be made on the rate and amount of moisture ingress at a given future point in time. The results obtained are in good agreement with current nonsaturated flow theory further supporting the validity of the proposed IRS ‘wick’ test. In a suitable soil, the ratio between the total specific surface area (SSA) of the aggregate fraction and the mass of the binder fraction appears to be positively
linked with the rate of capillary suction in rammed earth. When the mass of the binder fraction in a suitable soil is less than 10% of the total soil mass, it would appear that the rate of moisture ingress, due to capillary suction, in rammed earth is significantly increased. It is hypothesised, therefore, that through granular stabilisation (i.e. artificially modifying the particle-size distribution through additionysubtraction of material) the rate of capillary moisture ingress in rammed earth can be controlled. A simple initial suggestion may be to ensure a minimum binder quantity of approximately 10% by
Fig. 12. Particle-size distribution and IRS values for rammed earth soils with two parts binder.
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
279
Fig. 13. Particle-size distribution and IRS values for rammed earth soils with three parts binder.
Fig. 14. Relationship between IRS value and ‘3.35 ratio’ for rammed earth soils containing two and three parts binder.
mass. A ‘3.35 ratio’ of 5 has been proposed as a potential target value by which to attenuate the required approach to granular stabilisation. The 802 soil (worst), for example, can have its ‘3.35 ratio’ lowered from above 20.0 to below 5.0 by adding 12% by mass of 10 mm pea gravel. This should have the effect of lowering the rate of moisture ingress due to capillary suction, or even achieving an optimum value. This observation has significant practical applications but requires testing in the field in order to prove or disprove the hypothesis. Acknowledgments The author wishes to acknowledge the help of the following parties: Stephen Dobson (RAMTEC Pty Ltd)
for pictures, helpful advice and discussion Stephen Hetherington (Sheffield Hallam University) for technical support Paul Scholey (Rotherham Sand & Gravel Co Ltd.) for generous donation of earth materials. References w1x Hall M, Djerbib Y. Rammed earth sample production: context, recommendations and consistency. Constr Build Mater 2004;118(4):281 –286. w2x Dobson S. (RAMTEC Pty Ltd.) Pers. comm. w3x Oliver A. Dampness in buildings. Second ed. Oxford: Blackwells, 1997. w4x Building research establishment (BRE). The selection of natural building stone. London: HMSO, 1983. p. 1 –8 (Digest 269).
280
M. Hall, Y. Djerbib / Construction and Building Materials 18 (2004) 269–280
w5x Laycock EA. Frost degradation and weathering of the magnesian limestone building stone of the Yorkshire province. Ph.D. Thesis, University of Sheffield, 1997. w6x Hall C. Water movement in porous building materials – I: unsaturated flow theory and its applications. Build Environ 1977;12:117 –25. w7x Killip IR, Cheetham DW. The prevention of rain penetration through external walls and joints by means of pressure equalisation. Build Environ 1984;19(2):81 –91. w8x Camuffo D. Physical weathering of stones. Elsevier Science, 1995. p. 1 –14. w9x Vos BH, Tammes E. Flow of water in the liquid phase. Report no. B 1-68-38, Inst. TNO for building materials and building structures, Delft, Holland, 1968. w10x Kieslinger A. Feuchtigkeitsschaden ¨ an bauwerken. Zemet Beton 1957;9:1 –7. w11x Mamillan. Connaissances actuelles des problemes de remontees ´ d’eau par capillarite´ dans les mures. Rapporti della soprintendenza per I beni artistici e storica per le province di bologna, Ferrara, forli e ravenna. The conservation of stone II pre-prints of the contributions to the international symposium, Bologna,
w12x
w13x w14x
w15x
w16x w17x
w18x
27–30 October 1981: part A deterioration, Rossi-manaresi R. (ed.), Bologna, pp. 59–72. Vos BH, Tammes E. Moisture and moisture transfer in porous materials. Report no. BI-69-96, Nov 1969. Organisation for industrial research TNO, Institute TNO for building materials and building structures, Delft, Holland. British Standards Institute (BSI). BS 3921: 1985 (partially withdrawn) specification for clay bricks. BSI: London, 1985. Hall C. Water movement in porous building materials – IV: the initial surface absorption and the sorptivity. Build Environ 1981;16(3):117 –25. Hall C, Kam-Ming Tse T. Water movement in porous building materials – VII: the sorptivity of mortars. Build Environ 1986;21(2):113 –8. Minke G. Earth construction handbook: the building material earth in modern architecture. UK: WIT Press, 2000. De Freitas VP, Abrantes V, Crausse P. Moisture migration in building walls – analysis of the interface phenomena. Build Environ 1996;31(2):99 –108. British Standards Institute (BSI). BS 1377-2: 1990 soils for civil engineering purposes – part 2: classification. BSI: London, 1990.