1. Introduction The objective of this study is to clarify the blistering mechanism of membrane roofing applied directly on concrete substrate. In the previous paper, part 1, it was found that the main factors on the pressure increase to form a blister are moisture condition and air permeability of concrete. The moisture is the source of vapor pressure increase and the air permeability governs the pressure diffusion. Therefore, it is important to know both properties of concrete deck, and the practical influences of them on the pressure behavior. In this paper, a series of tests and a numerical analysis were conducted using concrete cured under various conditions simulated actual decks. 2. Concrete Specimens and Their Properties The mix proportion and the curing conditions of concrete specimens are shown in Table 1 and 2. The specimens were placed in the cylindrical PVC molds and cured in the molds for 28 days. After being cured, air permeability (Fig. 1(a)), moisture content and pore size distribution of specimens were measured. The air permeability depends entirely on the curing condition (Fig. 2), and it is physically determined by not only the moisture content but also the densification of concrete. The pore size distributions clearly show that the drying from early age obstructs the densification (Fig. 3). Concrete dried from early age has large pores to induce the extremely high air permeability. 3. Pressure Increase Test After the air permeability was measured, the concrete specimen was processed for the pressure increase test being made a void at the center on the top surface (Fig. l(b)). The void was imitated as a small chamber between roofing and concrete substrate to originate a blister. The specimens were put into a heat insulating wall between two environmental rooms (Fig. 4). One of them was controlled to provide a daily temperature change of 20℃-55℃-20℃ for the void, the other to be constant at 20℃, 40 % RH. The change curve of temperature and the maximum of 55℃ (Fig. 5) was supposed as a typical condition of roofing in a hot sunny day. The test was continued for 90 days and the daily pressure behavior in the void was observed by a manometer. When the test ended, the air permeability and the moisture content of the specimens were measured again. The solid lines in Fig. 6 are the results observed on the first five days and the last 90th day. It is clearly seen that the pressure increase is higher as the duration of sealed-curing is longer and the drying environment more humid. The pressure increase on the 90th day of every specimen became a little lower compared with on the first because the moisture content became a little lower and the air permeability a little higher. The high pressure increase, however, was kept in the specimen such as [4 w] or [1w-75]. It thus appears that a blister of membrane roofing grows through such a daily cyclic pressure behavior. 4. Relation Between Pressure Increase and Concrete Properties The relation between blistering and moisture content of substrate has often discussed in the literatures because more moisture contained in concrete can generate higher vapor pressure. From our results, the pressure increase became larger if the average moisture content was more than 6%, or the surface moisture content more than 5 % at the first period of the test (Fig. 7 and 8). At the last of the test, however, the relation between them was obviously different. On the other hand, the air permeability had a close relationship with the pressure increase in spite of the change of moisture content throughout the test period (Fig. 9). It is concluded that the air permeability of concrete has a significant effect on the pressure increase to form a blister, while the moisture content has a slight in practice. It can be said that a sufficient moist curing makes the air permeability low and the pressure increase high. It should be noted, however, that

(View PDF for the rest of the abstract.)