FIGURE 2. Apparatus for coating the backs of the specimens. Each specimen was placed face up on the top of the 11 -incylinder and the crank turned to apply a thin coating on its back. a. Smooth-Surface, Felt-Base Specimens The asphalts were melted in a double-boiler, consisting of a one-gallon can, wrapped in aluminum foil to prevent contamination, sitting in a twogallon bucket half full of asphalt. The temperatures of both the asphalt under investigation and the asphalt in the bath were watched constantly to prevent any overheating. When the temperature of the test asphalt was in the range of 400 to 450 °F (204 to 232 °C), any required mineral matter, preheated to 300 °F (149 °C), was stirred into it with a large chromium-plated ladle. The mixture was continually stirred at 450 °F (232 °C) until it no longer frothed. All compositions were made and are reported on a weight-percent basis. The saturated felts, cut into 9- X 12-in sheets, were heated in an air oven at 300 °F (149 °C) to warm them and to eliminate moisture. They were ready for use when the saturant, which came to the surface as the felts were heated, was completely reabsorbed. When the felts and coating were in working condition, a felt was removed from the oven and placed on a sheet of cellophane backed by a double thickness of 12-X 14-in Kraft paper and a 12- × 14-in piece of tempered hardboard. Approximately one and a half times the required quantity of coating was distributed over the felt and spacers of the required thickness (felt thickness plus 3 mils) were placed along two of the sides of the felt. A sheet of dextrin-coated paper, two sheets of Kraft paper and a piece of tempered hardboard were successively placed over the coating, and the assembly was put into the hydraulic press. The press was closed and the pressure raised to about 120 psig. After about 30 s, the panel was removed from between the sheets of Kraft paper and placed in a water bath at room temperature, where the cellophane and dextrin paper soon floated free. The panel was thoroughly washed and dried and its thickness measured. When the required number of acceptable panels had been prepared, the hot coating was ladled from the melting bath to the back-coater, which had previously been heated to about 450 °F (232 °C). While one man rotated the roller slowly, another passed the back of the panel over the roller, picking up a thin coating. The back coating was dusted with mica while still warm. Upon cooling, each specimen was cut to 7 x 11 in and exposed the next day on wooden roof decks, which had been covered with smooth-surface roll roofing (facing due south and at an angle of 45°). An area of 4 X 7 in was exposed, leaving each specimen a 3-in head lap. b. Granule-Surfaced, Felt-Base Specimens The asphalt and felts were prepared as in 2.4a. The hot plate (A) on the laboratory roofing machine (fig. 1) was heated to about 250 °F (121 °C) and the doctor bar (B) to 450 °F (232 °C). The canvas belt was threaded through the machine and a sheet of 12-in wide Kraft paper three feet long was fastened to its end, about 4 in in front of the doctor bar. In order to be assured that the doctor bar was at the proper distance from the machine bed, a blank run was made in which no granules were used. When cool, the thickness of the coated panel was measured. In a normal run the felt panel was fastened with masking tape to the Kraft paper on the hot plate. About one and a half times the required coating was distributed over the panel. The motor was started and the panel was pulled under the doctor bar (B) and granule hopper (C); then the motor was stopped for 2 min to permit the panel to cool and the granules to become firmly set. Then the panel was drawn through the S rolls (D) and embedding rolls (E). The panel was removed from the belt and passed backwards through the embedding rolls again. Because of the heavy gear on one side of the embedding rolls and because the rolls were not crowned, this procedure was necessary to make the embedment of the granules more uniform. The panels were backcoated as in 2.4a. The 47-in exposed portions of the specimens were examined through a transparent grid with 98 squares [0.40 × 0.40 in] covering the central 31⁄44 614-in area. The areas around the edges [0.325 in wide] of the 4- × 7-in area were excluded from the examination to eliminate the edge effects. The numbers of squares with cracks and spalled coating were counted and recorded during each inspection. Inspections were made every three months during the first two years of exposure and semiannually thereafter. The semiannual inspections were made in March and September. b. Aluminum-Base Specimens The aluminum-base specimens were examined with a high voltage probe in accordance with ASTM D 1670. Spark photographs were taken of all specimens with three or more breaks in the coating. These photographs were examined through a transparent 60- rectangle grid covering the central 2- 5-in portion of the specimen. Only failures by cracking were recorded for this type of specimens. 3. Progress of Deterioration 3.1. Smooth-Surface Specimens When the smooth-surface specimens were first prepared, they had a matte finish corresponding to the texture of the dextrin-coated release paper used on them. After short periods of exposure, the surfaces became glossy. Continued exposure produced elevated ridges, followed by thinner connecting ridges. These ridges were the result of expansion of the surface as it became oxidized. As oxidation progressed; the surface became progressively more water soluble; material was dissolved by rain outdoors or water spray in the weatherometers, and the glossy, ridged surface assumed a matte appearance. With time, widely spaced cracks appeared. These cracks usually penetrated the felt-base specimens fairly rapidly through the coating to the felts. In the aluminum-base specimens, the cracks frequently remained on the surface for relatively long periods of time and failure was evidenced by pinholing in noncracked areas or at crack intersections. In the felt-base specimens, widely spaced cracks became interlaced with narrower cracks until spalling of the small areas surrounded by cracks began. As coating spalled from the specimens, the substrates were exposed to the weather. In the aluminum-base specimens, exposure was usually discontinued prior to the spalling phase of weathering. In the felt-base specimens, the spalling permitted the saturated felt to be degraded rapidly by the weather. Photo oxidation rapidly destroyed the saturant and permitted water to enter the felts and, through alternate wetting and drying, shrink and distort them. Finally, the weakened tabs were blown away. 3.2. Granule-Surfaced Specimens The granule-surfaced specimens have shown only the early stages of deterioration during the 15 years they have been exposed to the weather. Changes observed early in the exposure period consisted of small losses of granules in most of the specimens and heavier losses of granules in the few specimens in which poor granule adhesion was measured prior to exposure. After about 12 years of exposure, cracks began to appear in the coatings. About a quarter of the specimens had cracks in 5 percent of their surface areas (five grid areas) after 12 and a half years. No spalling of the coatings has been observed in any of these specimens since their exposure in 1952. However, in the normal course of events, as weathering progresses, generally more granules will be lost, more cracking will occur and, finally, the coatings will spall off the saturated felts, just as in the smoothsurface, felt-base specimens. The granules protect Early in the exposure of the smooth-surface, feltbase specimens some small blisters were observed. These were principally around the edges, but in some of the specimens made with Mid-Continent and Venezuela asphalt, blisters appeared well within the areas inspected. Only a very few specimens made with California asphalt had any blisters; none had more than three. While blistering contributes to poor appearance, it does not interfere with the essential function of roofing, which is to keep the structures weatherproof. Therefore, blistering is not normally considered as a criterion of failure. However, blisters can act as centers of stress concentration and lead to earlier cracking and spalling of the coating. Thus, the effect of blistering is manifested in the reported cracking and spalling. The only other comment that can be made concerning blistering is that the largest incidence of blistering was associated with the coatings made from the more durable asphalts. The aluminum-base specimens and the granulesurfaced, felt-base specimens experienced essentially no blistering. 4. Failure Criteria When a product exposed to the weather no longer performs the function for which it was designed, it has failed. However, it has become customary to replace roofing for reasons other than, and earlier than, complete failures. Quite frequently, shingles are replaced because they no longer have an acceptable appearance; on other occasions failure can be anticipated in the near, but indefinite, future. Consequently, certain standards of failure were arbitrarily selected by the sponsors of this program. These failure criteria appear in table 5. 5. Outdoor Performance 5.1. Smooth-Surface, Felt-Base Specimens The results of 200 months (1623 years) of exposure of the smooth-surface, felt-base specimens to the weather are summarized in tables 6 and 7. Information on crack development is presented along with the observations of spalling. The ratio of the time in which an asphalt coating with mineral matter in it reaches a given level of deterioration to the time in which the same asphalt alone reaches that same level of deterioration is also reported. This ratio is frequently more significant than the actual number of months of exposure, for this comparison of a coating with mineral additives to its base asphalt tends to compensate for variations in the weather, or climate. This technique was used in reference [3], where the results of the accelerated weathering on these coatings were discussed. (A Spalls in 25 grid areas (25%). 50% Granule loss in 49 grid areas (50%) or spalls in 25 grid areas (25%). Cracks in 30 grid areas (50%) few of the coatings 15 mils thick (table 6) and many more of those 25 mils thick (table 7) had not reached some of the levels of deterioration reported. This condition is indicated by dashes in both tables). At every level of performance and in both thicknesses the Mid-Continent asphalt outperformed the Venezuela asphalt, which, in turn, outperformed the California asphalt. The mica-stabilized coatings out performed all the others; only one coating had deteriorated to any of the selected levels in tables 6 and 7 in 200 months. The 15-mil California coating with 35 percent mica exhibited its first spalled areas in 164 months. Before discussing specific performances, however, a word of explanation is in order concerning the mechanism of cracking and why cracking, though ultimately contributing significantly to shingle failure, cannot be used as an early warning of impending failure. While carly cracking of the coating on shingles may be startling to an uninformed observer, the data in tables 6 and 7 for smoothsurface specimens show the lack of correlation between early cracking and failure. Thus, cracking has not been used as one of the criteria of failure. Cracking of coatings on shingles results from the large differences in thermal coefficient of linearexpansion between the saturated felt and the shingle coatings. Data on these materials are difficult to obtain, but a few order-of-magnitude figures have been published. These are listed in table 8. It is evident from the above data asphalt moves roughly 25 times as much as the saturated felt does in the cross direction and 50 times as much as the saturated felt in the machine direction as the temperature changes. At the higher end of the range of temperatures encountered by roofing during exposure, the asphalt can accommodate the differential movements by flow, but at the lower end of the temperature range flow is not rapid enough and partial accommodation results by the curling of the shingle tabs. When this movement can no longer cope with the stresses, cracking occurs. These cracks usually run at right angles to the greatest differential movement. In the exposure of the smoothsurface specimens, all early cracking first occurred in the cross direction. These occurred very early in the exposure period, in some instances during the first winter, but the specimens continued to perform well for many years. There was no relationship between any of the levels of cracking and spalling. Therefore, nothing more will be said about cracking, for spalling has been set as the criterion for failure. Returning to the performances of specific systems, it can be seen that many of the specimens containing blue black slate have not deteriorated to the 25 percent spalled level after 17 years of exposure. For these systems, just as with the asphalts without |