# How to Take the Sample

The next question to address in the sampling is how to take the sample. Again, every building is unique and subject to specific circumstance, therefore the following example is offered for the purpose of illus­trating such distinctive characteristics and what may be done to address them: On a recent project the entire building exterior (13 stories) utilized fixed standing scaffolding. After interviewing the sealant applicator it was learned that a team of applicators working together in assembly line fashion (one cleaner, one backer rod installer and primer, and one sealant installer) had worked in a horizontal configuration around the building. The result was that the sampling was conducted vertically as an effort to sample potential changes in the crew’s application procedure over the installation time period. Historical infor­mation such as that obtained in this case is not ordinarily available, but the example is offered to illustrate the strategy one may use to obtain the most accurate sample.

If suspended power staging was the method used during construction, it is important to sample at least two separate areas to allow for possible variations resulting from different crews working on different locations. Since staging crews typically move the suspended power staging horizontally from one grid to the next, it can be assumed that sampling two grid areas side by side would not be as effective as sampling at completely different areas of the building. If knowledge of the construction access is not available, it nevertheless is logical to sample in the method suggested here.

Figure 3 shows as a further example an 18-story condominium project where balcony access allowed for comprehensive statistical sampling. In this case, balconies were accessed at random on all floors on every elevation.

 FIG. 3—Condominium project allowing balcony access for a comprehensive statistical sampling.

TABLE t—Statistics generated from sealant testing (Specimen A).

 Parameter Statistics Lincai- length of sealant joint 10668 m (35 (KM) feet) Number of adhesive failures 427 Average length of failure 50 mm (2 in.) Failure rate in linear length of joint one in 25 m (82 feet) Total linear length of sealant failure 21.6 m (71 feet) Percentage rate of sealant failure 0.2 % Percentage rate of sealant success 99.8 %

How Much Should be Sampled?

The next question is how much sampling should take place. Experience has demonstrated that if at least 5 % to 10 % of the building is sampled, the results are usually representative of the entire sealant installation.

For example, on an 11-story building, with 24 grid locations for suspended scaffolding, two grids were sampled. This represented an 8.3 % sampling rate. The inspection revealed 14 adhesion failures on one grid section, and 20 on the other, for an average of 17 failures per grid section. From this sample, it was predicted that 408 failures would be found on the entire building if testing was conducted at a 100 % rate. A complete 100 % inspection of the building revealed a grand total of 427 adhesive failures. This result indicates that statistical sampling forms a solid basis for obtaining a realistic view of the sealant installa­tion on a building.