Document RaO97zqwJNDOJrDM0NE9pD158

INTERNATIONAL ROOFING CONTRACTORS CONVENTION, BAD GODESBERG, WEST GERMANY JULY 1962 PLAINTIFF'S EXHIBIT NAR-698 INVESTIGATIONS OF MOISTURE PROBLEMS IN BUILT-UP BITUMINOUS ROOFS BY K. G. MARTIN COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANIZATION DIVISION OF BUILDING RESEARCH MELBOURNE, AUSTRALIA BIRD 012835 INTERNATIONAL ROOFING CONTRACTORS CONVENTION, HEIDELBERG, WEST GERMANY JULY 1962 INVESTIGATIONS OF MOISTURE PROBLEMS IN BUILT-UP BITUMINOUS ROOFS BY K. G. MARTIN COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANIZATION DIVISION OF BUILDING RESEARCH Y?!! MELBOURNE, AUSTRALIA BIRD 012836 INVESTIGATIONS Qg M3ISTURE PROBLEMS IN BUILT-UP BITUMINOUS ROOFS by K.G. Martin Division of Building Research, Commonwealth Scientific and Industrial ' Research Organization, Melbourne, Australia . I. INTRODUCTION Problems associated with built-up bituminous roofs in. Australia were first investigated ..by the Division of Building Research about' ten years ago; These- roofing membranes were-not giving reliable service and a survey of roofs throughout the country revealed three main reasons for this. Specifications to define the oorrect materials to use were not in existence; no details were available of how to design the membrane to form an integral waterproofing system over the. whole building, inoluding movement joints,-:parapets : and protrusions; and finally',: the heed to 'protect the waterproofing membrane'.with a weathering Surface maintained in good order was not-, appreciated. Work'commenced at the Division'by constructing experimental membranes at 'sites1 in the vicinity of Melbourne and closely observing their performance. Considerable knowledge was obtained from these' . investigations and formed the basis of a general study which was -' prepared to give design, data to-assist- the architect and builder -. (Ballantyne and Martin 1960). Laboratory examinations of roofing fabrics and bitumens* supplemented the work and as a result the Division was in a position to assist in the drawing up of specifications for bituminous roofing felts by the Standards Association of Australia (l 959, 1960). In these 'specifications both composition and performance tests were included, and in connection with development of the performance tests the deterioration of bituminous membranes that have no additional surface protection has been studied more closely.. Such . membranes are used in light low cost construction to roof dwellings, schools, libraries, And the' like, where the flat roof allows complex plans to be' easily roofed or where a flat, roof is required as an . architectural feature; ' The most common faults with these roofs; are v associated with' moisture- ingress. ' -This -paper summarizes recent Work' on.'bituminous roofing fabrics carried out -at -the Division and other basic' data which, contribute to an understanding of how the moisture problems-develop and how they may be minimized without resorting to higher' dost massive surface protection; - ' * The tejrm bitumen is .used in. a general sense embracing coal tar pitches, petroleum dr asphaltic bitumens and naturally occurring asphalts. BIRD 012837 II THE FAUHTS Within 1 2 months of exposure a membrane that consists of alternate layers of waterproofing bitumen and stabilizing bituminized fabric with only a coating of bitumen as a surfacing is likely to develop small bubbles in the coating, as shown in Figure 1. These may remain the same size far many years, or they may burst and leave holes through the coating. The membrane itself may also develop much, larger bubbles comnonly known as blisters; a particularly bad case of which is illustrated' in Figure 2. After about 5 years without maintenance the top layer, of fabric may shrink and distort or become extremely weak and-rupture. This type of deterioration is illustrated in Figure 3- One of the most inportant factors is the composition of the bituminized fabric. . Both saturated and coated fabrics are used for constructing built-up roofs in Australia. The former are those produced by impregnating a raw fabric with a low softening point bitumen and squeezing out the excess so that the material retains a dry hairy surface. A coating of a high softening point bitumen applied to the saturated fabric converts it to a coated fabric, which is dusted to prevent sticking when the product is made up into rolls. The raw fabrics used are cellulosic felt, asbestos felt, woven hessian, or glass fibre random tissue, the latter differing' from the others in that the fabric is saturated and coated with the harder bitumen in one operation. III. SOME BASIC STUDIES (a) Water absorption The layer of bitumen providing the surface .coating is usually 0.0) to 0.04 in. thick, and with one surface in contact with water will slowly absorb moisture at a continuously decreasing rate. The study of this process is difficult because absorption apparently ceases after about 100 days at normal temperature, when solution of the bitumen by water and volatilization of the bitumen (particularly coal tar pitches) counters fee increase in weight due to water absorption. Under these circumstances various bitumens have been compared by fee weight of water absorbed during an initial speoifio period per unit of exposed surface area. It has been found that numerous factors such as fee type and hardness of the bitumen and the presence of salts and BIRD 012838 3 fillers influence the rate of water absorption, "but the effects are not important in gelation to the observed, faults since in all cases the rate of water absorption is extremely slow (of the order of 0.0001 Ib/sq.ft./day maximum initial rate) . The absorption of water into bituminized fabrics has mere important effects, particularly in the saturated cellulosic felts which absorb water about 500 times faster than the bitumens. The absorbed water causes an increase in the thickness of the organic fibres, and since the fibres are orientated more in the machine direction (length of roll) than-in the cross direction, most swelling of the fabric takes place in the cross direction. Another factor inherent in raw felt manufacture adds to this effect. Part of the shrinkage in the raw felt while it is passing over the drying rollers is prevented by the tension that is applied in the machine direction. When the felt is subsequently rewetted this prestrained condition counterbalances some of the swelling in this direction so that again most of the movement takes place in the cross direction. The saturating bitumen does not prevent water absorption and subsequent swelling because the bitumen enters only 75 per cent, of the volume of voids in cellulosic felts and only 55 per cent, in asbestos felts (Brown i960). The asbestos felts con tain about 15 per cent, of organic fibres' in order to give good felt, formation and consequently these fibres will also swell when in contact with water, but the overall swelling of asbestos felts is much less. Glass fibre random tissue relies upon organic adhesives to provide - inter fibre bonding and moisture may slowly displace the adhesives from the raw tissue. . However, bitumen impregnation of the very open tissue eliminates practically all voids and only a negligible amount of moisture is absorbed by glass fibre roofing material. On drying out the cellulosic fibres shrink, but this does not take place uniformly and redried felts are thicker and permanently shrunk, particularly in the cross direction. The amount of permanent shrinkage varies with different felts and has been compared by sub jecting them to four cycles of immersion in water and oven drying and measuring the change in the cross direction dimension from the initial : wet to the final dry condition (Martin 1959)* Some shrinkages of' bitumen-saturated fabrios in this test are: Cellulosic felt, poor quality Cellulosic felt, good quality HessianAsbestos felt Class tissue (coated) 2.5 per cent. 1.5 1.5 0.2 " " Nil It has also been noted that the moisture absorbed by the bitumen-saturated felts releases inter-fibre bonds and considerably reduces the tensile strength. On, drying out some bonds do not reform, and in this way the tensile strength of roofing felts BIRD 012839 4 deteriorates as a result of .outdoor exposure (Martin 1 961) 4 comparison of this behaviour .for various bituminous saturated felts has been made by measuring the tengile strength of some fabrics, after cycles of immersion in water at 77 P for 8 hr and oven drying at 130-440 r for 16 hr. Results obtained on 1. in. wide strips by testing at 1 in. per min, over-a-3-in. gauge, length, are given in Table 1 . tabt.f: i . TETEILE STRENGTHS OP SATURATED PELTS Material \ Mean Tensile Strength (lb) Fabric Type JjXliUuj&U-l Saturant ... Dry No ' 5. Treatment - Cycles Wet 15 . 2 - Circles Cycle. Glass Asbestos Cellulosic Celluloslc (English) Hessian Blown asphaltic Residual asphaltic Residual asphaltic Goal tar 23 : ^' 26 45 , ?6' . 20 .-22 37 27.5 17.5 20 ' 22.5 Residual -asphaltic . 9.2 -- ' ` 9.5 . 9.0 24 12 9 12 6.5 These results indicate a marked superiority of glass fibre fabrics, but this applies only when-moisture has penetrated, to the . fabric fibres'themselves. When comEercially available. coated felts. are examined it is found that considerable moisture is absorbed, the- ` i. water absorption varying between different samples. of. roofing but often . being ohe-fifth of. the amount absorbed by the saturated felts. The." moisture is stored in the saturated felt portion^of the fabric,.and the high water absorptions are due either to a lack of integrity of the coatings or to a high rate of diffusion through the coatings. (b) Moisture permeability Moisture, transfer through intact bitumen coatings is a well understood process which may be described by mathematical formulae involving the concept of a moisture permeability constant (Martin 1958). However, these, formulae, apply only to what is known as the steady state condition, which takes of.the order of 10 days to become established, across o.02 in. thick, layers of roofing grade bitumens kept at a con-.. stant room temperature and a particular relative humidity. Bitumens of ~\ BIRD 012840 5 similar hardness have, similar moisture. permeability constants, which . are extremely low*.. . A blown.'asphaltic MtumeniWill incrgase in permeance about five times over the temperature range 40 to-i40T?, and. although coal tar pitches'.and residual asphaltic bitumens would increase . more than this- the effect is not very important when, the low orders of . permeance are-considered*. . . . Bituminous-coatings to roofing felts usually contain inorganic fillers which have been found to reduce the permeance of intact coatings, more moisture being stored in a filled coating but less penetrating it. The permeance of commercial coated felts should he similar to. that of an intact bitumen coating of similar thickness to the coating on' the felt. However, it is found that the coated felts are mary times more permeable than this, as is. shown by.the results given in Table 2. TABLE 2 TIPICAL RESULTS FOR J40ISTURE PEBMEANCE Sample Thickness (in.) Permeance (perms)* 70F 1 20F 00 *C* Asbestos felt Saturated - 0.026 4.0 4.0 4.6 ,4.0 Cellulosic felt. Saturated, 0.046 10.7 12.1 10.2. 1.0.5 . Asbestos felt . Coated 0.063 0.27 0.12 0.14 0.11 Cellulosic felt. Coated 0.067 0.23 . 0.15 0.15 0.17 Glass Fibre Roofing1 0.05. ,0.06 0.04. 0.07, ..Q.ca: Laboratory made Coating 0.020 0.003 0.005 0.009 0. ot 4 * 1 perm is 1 gr of water transferred through 1 sq.ft, of material in 1 hr under a vapour pressure gradient equal to a head of i in. of mercury. BIRD 012841 6 The coated felts "behave similarly to the saturated felts in.that they are little influenced by temperature except for an increase of . permeance at the lowest temperature.. It is concluded' from these results that coatings to conniercially available felts contain micrccracks which become larger at lower temperatures as a result of. thermal contraction of the coating (the cubical coefficient of thermal exapansion for . bitumen is about 3x1 cHyF, which is ten times greater than for most metals). Migration of water through the holes is obviously more important than diffusion, of moisture through the intact coating. If three or more layers of coated felt .(or saturated fabric) are completely bonded together with layers of bitumen to form a membrane,the transfer of moisture through the membrane takes years to reach the steady state. Attests to measure the permeance of these menbranes have been unsuccessful, but it has been established that' with liquid water on one side and an atmosphere of low relative humidity on the other no moisture transfer takes place during the first 50 days. With regard to the problems tinder discussion .the membranes as a whole may be considered impermeable. Some moisture will enter the exposed surface of a newly constructed roofing membrane and may penetrate into the top layer of reinforcing fabric, but moisture will not pass through the membrane, unless it ha&.ruptured...................................... (c) Pressures developed by confined water vapour and air Built-up bituminous membranes may contain air trapped in two ways; firstly, the roofing fabric may contain voids, and secondly, voids will occur where imperfect adhesion occurs between the bonding bitumen and the deck or the roofing fabric. It is of interest to note how little moisture is required to saturate air voids under temper atures comparable to those existing in roof membranes, viz; and 0.00041 Ib/ftl at 4-05? 0.00158 Ib/fti at 80T O.OOA92 lb/ft, at 1 2031 0.0129 lb/ftJ at 150~F - - Considerable increases of pressure will develop within these voids if they are restricted to a constant volume and simultaneously heated. Table 3 shows the pressure increases due to heating a confined void initially in equilibrium with external air pressure at 70 S'. > V BIRD 012842 7 TABLE 3 'in-*---: HRESSUHE IIKJSEASES' IN' VOIDS Content of Voids Pressure Increase`1. . for. Temperature Reached. (lb/sq.in.) 110% 130F 150% 1. Gry air 2.- Air and water vapoug^j saturated 1.1 1.1 1.6 1.6 2.2 2.2 3. Air, water vapour, and liquid water 2.0. 3.4 5.5 The presence of sufficient water to maintain the relative humidity of the void at 100. per cent, over the temperature range involved causes, the niftyTiwim pressures to be developed. A roof reaches much higher temper atures than the surrounding air a^d. it has been shewn that daily temperature cycles of 70 F to 150 F may occur many times during exposures in Australia, unless precautions are taken to reduce membrane temper atures by applying.reflective surfacng3 (Ballantyne and Spencer 1957). Interesting findings have been' made concerning the voids in the roofing felts. ' Fresh samples of coated cellulosic and asbestos . felts direct from the manufacturers have been found to contain about 1.3 and 0.6 per cent, of moisture respectively (Brown i960). This moisture is derived from the inherent hygroscopicity of the raw felt, and this property has been investigated. The moisture content of typical raw felts has been found to depend upon the relative humidity and temperature of the air surrounding the felt, as shown in Figure 4. It is important to note that the moisture content decreases with increase of temperature, indicating that moisture held by the felt at ordinary - temperatures becomes free at higher temperatures. The moisture content, left in coated felts after processing is 8.8 per-cent, of the content .of raw cellulosic' felt and 2.8 per cent, of the raw' asbestos felt. . According to the results'.in Figure 4 this means a relative " humidity in the voids of' the coated felts of a little more than >- 50 per cent, at 70%, but more than enough free moisture to saturate the voids at temperatures exceeding 1 00 F. It follows that the maximum pressures develop in these voids when the coated felts are exposed to solar radiation. BIRD 012843 8 (d) Thermoplasticity of bitumens Each of the faults to be discussed involves a deformation which overcomes the viscous resistance* of either' the bonding or coating bitumen. Thi3 resistance to deformation or viscosity depends upon the grade or hardness of the bitumen and varies most markedly with temperature, as shown in Table 4. TABLE 4 VISCOSITIES. OP. BITUMENS Type Blown asphaltic Blown asphaltic Residual asphaltic Residual asphaltic Coal tar pitch. Softening 215 185 145 135 . .135 Hardness (penetration at 77%) (O.OI cm) 15 30 25 45 . 15' Viscosity (poise) at 32% 9 2x108 IxtO9 5x10 8x105 77^ 150% -g- 1x10 1x10 2x10 2x1 07 ixl 03 1x10^ 5x|0 2x1 07 5x103 4x10* The bitumens approach similar viscosities at very low temperatures but at normal and higher temperatures the coal tar pitches and residual asphaltic decrease in viscosity much more than the blown asphaltic bitumens. IV. FAULT MECHANISMS (a), Small, bubbles The; small bubbles are most commonly found on bitumencoatings over cellulosic or asbestos base felts, but rarely over glass fibre roofing fabrics, which absorb a.negligible amount of moisture. The base of the deformation is at the interface between the coating *\ BIRD 012844 9 and the -saturated fabric, and it is believed that the deformation is. due to. the. expansion of- the-, moisture-saturated air voids in .the' ' . saturated; felt portion. As previously discussed, rain water .mhy. also penetrate inicrocracks in. factory-applied coatings and. enter the felt. This usually takes place at comparatively low temperatures'/;' the ' moisture migrates .within the felt whi-le. the. pressure in.the voids remains approximately in equilibrium with external air pressure. When the sun heats, the coated felts temperatures of more than 1 001? are reached in a few hours, and the pressure in voids that.;do not gain relief by venting through microholes, increases (as in case 3', Table 3) Simultaneously- the coating softens (as shown in Table 4-) and eventually deformation may take place at higher temperatures. - Upon cooling, the viscosity of the coating increases and offers more resistance to any. suction that develops as the vapour pressure of the void decreases. So that ary deformation is likely to be permanent. Cycles of heating and cooling may thus cause the deformation to increase until the increase in volume of the voids sufficiently compensates for the increase of pressure, or until the coating bursts. Another factor is that the coating progressively hardens as a result of weathering and this retards progressive growth of the bubble. (b) Shrinkage and puckering -Shrinkage and puckering of the top layers of roofing fabrics- are now easily explained as being the dimensional movement induced in the fabric by cycles of wetting and drying. After pro longed exposure, holes, .in the surface coating become more numerous as the pustules break and cracks develop in the. age-hardened coatings. Water absorption by the fabric and also drying out become more rapid and the moisture content of the fabric.follows .cycles of wet and dry weather. The usual conditions of cool wet and warm dry weather mean that the swelling of the fabric tends to be restrained by the more viscous bitumen but the shrinkage is permitted by the less viscous bitumen; the net. result is shrinkage which eventually distorts and puckers the fabric. ' The shrinkage of saturated fabrics by this, mechanism may be as much as 20 per cent (Martin 1959) ? 'which is about ten times that of the saturated fabrics loose laid (not bonded with bitumen). (c) Blistering * < ..... The .mechanism of blistering is simi lar to that of bubble formation except that the voids are due to imperfect adhesion, of the bonding bitumen to the deck or to the bituminized fabric. The presence of moisture in the void usually results from the use. of inherently wet materials (e.g. concrete decks) or it may be from rain which , fell, during- construction.. Concrete provides a particular hazard because . the surface may look dry even though little of the free water.lias, been removed. . This is because the drying out depends upon two processes. BIRD 012845 10 . ' I" ' * .' ` im the diffusion, of .moisture from the interior to the surface and the evaporation of the moisture from.the "surface, and. of these the first process.is very much, slower than the second. .Attests "to bond bitumen to a damp surface give areas of "no adhesion and provide air voids saturated .with water vapour. Water may also slowly displace the bonding bitumen from an adhering surface, and enlarge the void. Moisture has been observed to gain access at a faulty membrane turn up and to migrate in a channel between layers of fabric to an area of poor adhesion, where a large blister has formed. Poor adhesion may also be caused by inisses in the bonding bitumen or areas where the bonding bitumen has not wetted the surface. The talc or ndca dusting an coated felts makes it more difficult to wet the fabric with the bonding bitumen. V. HOW TO REDUCE THE INCIDENCE OF FAULTS It is now possible to suggest ways to improve the performance of roof membranes without using costly massive surfaceprotecting layers. Surface coating bubbles will be minimized if a fabric which absorbs a negligible amount of moisture (glass fibre based roofing) is used as the top layer of the membrane. If other fabrics (coated felt3, saturated asbestos felt or saturated hessian) are used the resistance to deformation of the final surface coating should be as high as possible. Thi3 can be achieved by applying a hard grade of bitumen in the form of a paint to the top layer of fabric after it has been laid. Finally, the peak temperatures of the membrane should be reduced as much as possible to minimize any build up of pressure. This means that the membrane must have a finishing coat of light coloured paint which has a low film strength and will nob damage the underlying bitumen. Aluminium leafing paints with bitumen vehicle and some FFA emulsions have been found suitable but are not particularly durable themselves. However, this is not a disadvantage since the need to repaint after about 2 years provides an excellent opportunity to inspect the condition, of the membrane and to repair ary minor faults. At this stage a more durable higher film strength paint can be applied (e.g. a varnish-based aluminium leafing paint). After this, further treatment will be necessary at only 5 to 10-year intervals. The same procedures will reduce shrinkage and puckering; a top layer of glass fibre fabric because it shows no moisture movement, an in-situ applied surface coating of hard bitumen kept in good con dition to reduce the rate of moisture ingress, and a reflective surface treatment to lower peak temperatures and minimize the changes in viscosity of the bitumen bonding the fabric. n BIRD 012846 11 ' : The ways of improving blistering L are not as straightforward because much depends upon the circumstances under which roofing'. . operations are carried out, and on the workmanship. Blistering can be reduced-by excluding, the-voids and by applying a reflective surfacing.^ to the-menibrane-. Concerning the practical problem, of exclucling .'voids during construction of. the membrane the most troublesome enemy, is. moisture. Bituminous roofing should be constructed only,during dry weather and the surfaces concerned should be a3 dry as possible. In inherently damp decks the simplest. way to overcome .sub-membrane blister ing is to lay the membrane'loose with sufficient, fixing to resist wind suctions. Any pressure build-up is then distributed over a large, area of deck and some relief may be obtained by providing vents from the deck to-outside air, either at skirtings or by means of. small stacks in .the- centre of the roof. If rain wets bituminized felts-during laying operations the only way to minimize intermembrane blistering is to discontinue work`until the . surfaces are-dry. Sometimes roofing must continue during suspect weather and if this is- likely it is advisable to use the fabrics that absorb the least amount of moisture. Other factors that militate against, good adhesion are the presence, of loose dust on the surfaces to be bonded, the use of bitumen that is too cool and consequently too viscous to wet- the surface properly, and attempts to bond to surfaces that- are not flat with fabrics that do not have good pliability. Although proper attention to workmanship may overcome these difficulties the situation can be simplified by using fabrics that are not heavily dusted and which have good pliability characteristics. It should be noted that this latter requirement infers the use of the thinnest fabrics (Ballantyne and Martin 1 958). The appropriate tenperature for application of the bitumen can be predicted (Martin 1960) and should always be used; finally membrane details should be designed to be as simple as possible. V. REFERENCES Ballantyne, E.R., and Martin, K.G-. (1958).- Pliability testing of bituminous coated roofing felts. C.S.I.R.O. Aust.Div.Build.Res. Rep. 02.1-6. Ballantyne, E.R., and Martin, K.Q-. (l 960).-Bituminous roofs. C.S.I.R.O. Aust. Div. Build. Res. Build. Study No. 1. Ballantyne, E.R., and Spencer, J.W* (1957). - Temperatures of bituminous roof membranes. Bull. Amer. Soc, Test. Mat. 1 957 (223) 169. BIRD 012847 12 Brown, NiG.' :(V960^.-The'analysis of .bituminous felts-,- C.&.I.R.O. Aust. Div. 'BuilcU Rds* Rep. 02.-1 -^8. .. . Martin.; K<G. (l'95S)V^ The addition, of inorganic, fillers ,to.bituminous coatings forroofing. 2-.- Moisture-pera^ability .oflrituminaus coatings and roofing felts. C.S.I.R.Q. Aust. Div^,Build. lies. ... "Rep. 02.5-6. . Martini K.G. (i 959)*- Changes in Bituminous roofing felts. associated with changes in moisture content. C.S.I.R.O. Aust. Div. Build, Res. fedi. Pap. lfo.8;-. : ?: . ... Martini (l 960)- Bitumens for Built up roofs. -1* Specification testa. C.S.I.R.O. Aust. Div. Build. Res. Rep. 02.5-8.- Martin, /K.G, (i 9&\ )Deterioration of. Bituminous roofing fabrics... C.S.I.R.O. Aust. Div. Build. Res. Tech. Pap.- No,: 11. -- ;. Standards Association of Australia (l 959) - Specifications for Bituminous Pelt Roofings; . . A.S. No, ,98 Type 1 (a) saturated organic fibre felts; A.S. No. 99 Type' l (c) coated organic fibre felts. Standards Association of Australia (l 960). - Specifications for Bituminous Pelt Roofings:' A.S. No.-120 Type 2(a) saturated asbestos fibre felts; A.S. No.1 2\ Type 2(b) coated asbestos fibre felts. BIRD 012848 FIG. t - SURFACE BUBBLES ON AN AGED BITUMEN COATING C* BIRD 012849 FIG. 2 - BAD BLISTER IN G IN A F LA T ROOF M EM BRANE BIRD 012850 FIG. 3 - CROSS DIRECTION SHRINKAGE AND DETERIORATION OF CELLULOSIC BASED ROOFING FELT BIRD 012851 MOISTURE CONTENT ( PER CENT. ) RELATIVE HUMIDITY ( PER CENT. ) FIG.4 - THE HYGROSCOPICITY OF RAW ROOFING FELTS ( MOISTURE ADSORPTION ISOTHERMS ) BIRD 012852