WEB EXCLUSIVE: Moisture-related failures in agglomerated floor tiles

Measuring curl across the tile.
November 08, 2011

Of the countless innovations related to technology, means and methods of construction, and product development, one notable example is agglomerated stone tiles. Agglomerated stone is used for many applications, such as countertops, tables, wall cladding, and flooring; it is an artificial stone product that is replacing more expensive products such as natural stone or terrazzo. Agglomerated stone tiles have an attractive appearance; they can easily be mistaken for natural stone marble or granite, while actually they are an engineered, man-made material."Agglomerated" means a group or mass of objects loosely thrown or huddled together or gathered into a ball or cluster, such as a mass of volcanic fragments linked through the action of heat. Similarly, agglomerated stones are engineered and manufactured by uniformly mixing natural stone fragments (e.g., marble, quartz, or granite) with a binding material.

The manufacturing process consists of crushing natural stone fragments, in a variety of colors, into different-sized pieces of aggregate, from a powder to large pieces. These aggregates are then mixed with a polyester resin, epoxy matrix, or in some cases cement or lime (also called terrazzo tile). Often, when cement is the binder, the manufacturing includes a suction process that extracts water from the paste, thus leading to a very low ratio of water to cement. The different mixtures are placed in a mold, in many cases project-specific, and transferred into a vacuum chamber, where the aggregate is vibrated and compressed to eliminate voids in the stone mix. This manufacturing process is typically performed in a room-temperature environment. Polyester resin and epoxy bound tiles cure through polymerization, while cement-bound tiles cure through hydration. 
 
As noted above, agglomerated stone was developed as a more cost-effective alternative to natural stone. In addition, these stones can be designed and customized to the end user's needs--shape, thickness, hue, and shine. They also have the advantage of being manufactured to have a consistent look (as opposed to the sometimes unwanted variations that can occur with natural stone, but which give stone its appeal). They can be engineered to have better performance than natural stone in terms of stain resistance, improved flexural strength, and scratch resistance. Agglomerated tiles also have an advantage over natural stone in that they make use of recycled or otherwise unusable stone scraps and transform them into a high-end finish material. Using recycled material is an important criterion for many end users and for LEED certification.
 
Agglomerated tiles have their limitations, however. As with many newer materials, these limitations are often discovered and understood when failures of in-service products occur and an investigation follows. We have observed two main failure mechanisms with agglomerated floor tiles: curling and cracking.
 
Every agglomerated stone tile has some degree of dimensional instability due to moisture. The amount of this instability varies from tile to tile. The Gabrielli test, a method developed by Mapei International Co. in Italy and outlined in the British Standard EN 14617 -12:2005 Part 12, can be used to measure the amount of this instability. The test measures tile displacement over time at specified points on the tile while the back side of the tile is exposed to liquid water. The result of the Gabrielli test then gives considerations for the installation of the tile. 
 
The tiles are neither ceramic nor stone, both of which have installation guidelines written by their respective industry groups (e.g., Tile Council of North America Installation Guidelines and Marble Institute of America Design Manual). These guidelines, as well as several ANSI standards, are the only references for the installation of agglomerated tiles, though they were not specifically written for the product. Right now there are two methods of installing agglomerated stone tiles for an interior flooring assembly. The first method uses a cement mortar (sand, cement, and water) to bond the tiles to the substrate. Tile installations using cement mortar introduce water immediately to the tile bottom. As the mortar and the concrete slab below cure, water vapor is released and travels up slowly through the tiles (which may have a low permeability) and more quickly through the mortar joints between tiles (which have a higher permeability than the tiles). Tile size factors into how quickly the water vapor from the mortar and concrete is able to escape up through the joints. When large tiles are used more moisture is trapped below the tiles. This is because there is more tile surface area and less tile joint area.
 
The second installation method uses an epoxy or other resinous adhesive mortar to bond the tiles to the substrate. Epoxies are typically a two-part chemical mixture and are high in bond strength. One of the major advantages to using an epoxy is that it undergoes a chemical cure which does not introduce water to the agglomerated stone tile assembly during the curing process.
 
Agglomerated tiles are typically installed directly over concrete or over a membrane above the concrete. These membranes may be intended as sound absorbing, anti-fracture or crack isolation, and waterproofing. For slabs-on-grade, an additional source of water vapor beyond the concrete itself may be from moisture in the soil. Many modern slabs-on-grade are built with a vapor retarder below the slab to impede moisture vapor from the soil from infiltrating the concrete slab assemblies. Older slabs-on-grade often have no vapor retarder. A replacement of older stone flooring, which has performed adequately for decades on a slab without a vapor retarder, with agglomerated tiles may lead to a series of problems because the new tiles are less permeable and trap the water vapor that used to dissipate through the natural stone.
 
We had the opportunity to investigate a large-scale agglomerated tile flooring failure in a new medical facility in Northern California. The owner had installed agglomerated tile throughout the first-floor corridors. Less than a year after they were installed, the tiles had curled to such an extent that they not only created an aesthetic concern, but were also a significant safety concern, making the floor basically unusable for the occupants of the building, in particular, for wheelchairs and gurneys. During our preliminary site observations, we measured across the 24-inch tile width an average tile curl of 0.08 inches and a maximum tile curl of 0.16 in. (~5/32 in.). 
 
 
The flooring assembly on the project consisted of the following (from the bottom up): 
  1. Lightweight concrete over steel decking (concrete thicknesses varied from 4.5 in. to 13 in.
  2. Cementitious self-leveling underlayment 
  3. Adhesive
  4. Crack isolation sheet
  5. Thin set latex cement mortar
  6. 24x24x1/2-in. agglomerated tiles made up of marble chips and polyester resin. 
We performed some qualitative nondestructive testing on the tiles. The simplest test was dropping a half-inch-diameter ball bearing on the tiles. By comparing the sounds and bouncing height after the ball bearing impacted the tiles, we could tell whether the tile was well bonded to the substrate. The solid sounds in the center of the tile were clearly different from the hollow sounds at the edges where we observed the curling; further, the balls bounced higher on the solid areas than on the hollow areas. The curling action of the tile is associated with dimensional instability (i.e., expansion of the bottom surface relative to the top surface). Because the temperatures below and above the flooring system were fairly stable we were able to rule out thermal expansion as the cause for the curling, which left moisture-related expansion as the probable cause.
 
We conducted several field moisture tests on the floor assembly to better understand the site conditions below the finished floor. These included moisture vapor emission rate testing (MVER) of the cementitious underlayment and concrete slab in accordance with ASTM F 1869, Standard Test Method for Measuring Water Vapor Emission Rate of Concrete Subfloor Using Anhydrous Calcium Chloride, and relative humidity tests of the concrete slab in accordance with ASTM F 2170, Standard Test Method for Determining Relative Humidity in Concrete Floor Slabs Using in situ Probes. We found that the cementitious underlayment and concrete slab had an average emission rate of 8.3 lb/24 h/1,000 sf and 9.6 lb/24 hr/1,000 sf, respectively.
 
Both of these MVER measurements are significantly above the industry standard, which is typically 3 lb/24 hr/1,000 sf for highly impermeable and sensitive flooring, while 5 lb/24 hr/1,000 sf is the typical maximum accepted MVER for most permeable flooring products. We also measured the relative humidity within the slab and found that it was between 85% and 90% RH. ASTM F 710, Standard Practice for Preparing Concrete Floors to Receive Resilient Flooring, and many flooring manufacturers recommend that the relative humidity within the concrete be 75% or less before installing the flooring products. The agglomerated tile manufacturer for this project gave no information on MVER or relative humidity limits for its products; however, the crack isolation sheet adhesive manufacturer limited MVER to 5 lb/24 hr/1,000 sf. These tests confirmed our concerns about the presence of elevated levels of moisture beneath the tiles from the concrete slab and leveling underlayment. Because the concrete slab was on a non-vented metal deck (which acts as a vapor retarder), the slab dries primarily from the top side. Lightweight concrete takes a long time to dry from one side of the slab only. 
 
This floor assembly used a crack isolation sheet, a composite sheet manufactured from chlorinated polyethylene (CPE), which was adhered over the self-leveling underlayment. Due to the elevated moisture from the concrete slab, the adhesive was tacky, had lost its adhesive properties, and caused loss of bond between the self-leveling compound and the crack isolation sheet. The tiles and mortar were adhered to the crack isolation sheet. The loss of adhesion strength of the crack isolation sheet to the self-leveling compound allowed the tiles to curl. Many of the tile corners that curled later cracked from foot and wheel traffic. If the tiles had been fully attached to the substrate below and tended to curl, the buildup in stress across the tile could have cracked the corners of the tiles, as we have seen on a different project. 
 
Crack isolation sheets are often laid and attached over concrete slabs to prevent cracks in the concrete from translating to the mortar and cracking the tile above. The action of the crack isolation sheets is in the horizontal plane of the floor tiles. These sheets are not intended to restrain floor tiles in the vertical direction. Floor assemblies that include crack isolation or sound isolation sheets should be reviewed to determine the attachment strength of the sheet in a direction perpendicular to its plane.
 
Next, we needed to better understand the physical properties of the tiles themselves. We tested the tiles for absorption, average density, and permeability (i.e., ASTM C 97, Standard Test Method for Absorption and Bulk Specific Gravity of Dimension Stone, and ASTM E 96, Standard Test Method for Water Vapor Transmission of Material, respectively). While we found that the density was comparable to the manufacturer's published values, the absorption rate was significantly higher (0.11% by weight as compared to the manufacturer's reported maximum of 0.04% by weight). With the permeability test, we found that the tile has a low permeability (0.02-0.03 perms) and effectively acts like an impermeable vapor retarder (<1 US perm is an impermeable vapor retarder). We also tested the crack isolation sheet for permeability and measured an average of 0.16 perms, which was about three times the quoted value given by the manufacturer of the sheet. These results show that the permeability of the crack isolation sheet is higher than the permeability of the tile. Therefore, moisture from the concrete will get trapped between the crack isolation sheet and the tile above. This trapped moisture induced curling of the tile; the curling was not restrained because of the loss of bond between the crack isolation sheet and its substrate. 
 
To further investigate the tile curl, we performed the Gabrielli test for determining dimensional stability by measuring tile displacement over time. The displacements are measured in two directions at different locations on the tile, such as the tile center, tile corner, and center of tile edge. These displacements are then used to estimate the tile expansion and curl. This test method exposes the back side of the tile to liquid water. We also wanted to simulate the conditions occurring at the project site and exposed the back side of the tile to water vapor. We called this test the Modified Gabrielli test We found that the Modified Gabrielli test with water vapor resulted in a greater curling of the tile than the Gabrielli test with liquid water. Figure 1 is a tile displacement vs. time curve that we plotted for a Gabrielli test. Figure 2 is a tile displacement vs. time curve that we plotted for a Modified Gabrielli test. Currently ASTM has no standard test that is equivalent to the Gabrielli test. 
 
The Gabrielli and Modified Gabrielli testing confirmed that the tested agglomerated stone tiles were dimensionally unstable when exposed to liquid water and water vapor. In the case of our investigation, the tiles probably curled first from the water in the thin-set mortar and later from the moisture vapor emissions from the concrete slab. During the investigation, we removed a sample of curled tiles from the building, took curl measurements over time in our laboratory, and found that the tiles kept an average of approximately 64% of their curl after one year of observation. This means that the curl was only partially reversible; even if the vapor emissions from the concrete had stopped at some point in the future, the curl would have remained. Polyester resins have large moisture expansion movement properties. Resins typically do not behave like linear elastic materials. As a result, deformations which occur due to moisture do not completely reverse when the moisture source is removed.
 
RECOMMENDATIONS FOR USING AGGLOMERATED TILES
Agglomerated tiles may be considered as an alternative to natural stone and terrazzo. Our findings should not overshadow the benefits with respect to economics, aesthetics, and performance of using agglomerated tiles installed properly over concrete with acceptable moisture levels. Rather, they should help our industry understand how these products perform so that professionals can make more informed product selection, design, and installation decisions. Specifically, we suggest the following: 
  • Limit moisture vapor emissions and moisture in the concrete slab. Perform the necessary concrete moisture tests to quantitatively understand the moisture levels in the different materials of the floor assembly. If elevated moisture is found in the concrete slab, a topical moisture vapor mitigation system should be considered prior to installing any floor finishes
  • Develop standard industry tests to allow manufacturers to report propensity to curling in a uniform manner. Together with the tests, develop moisture limits for the concrete substrate and curling limits for the tiles below which they will have acceptable performance
  • Develop standard industry installation guidelines specifically for agglomerated tiles. Conduct studies using mockups to validate acceptable performance for floor installations
  • Limit moisture related to construction. Carefully select mortar and grout products (follow the mortar and grout manufacturers' recommendations on what products to use for highly moisture sensitive agglomerated stone)
  • Limit underslab moisture. Moisture accumulation below slabs-on-grade is a common phenomenon due to moisture and vapor drive from the soils below. It is industry standard to install a vapor retarder below a slab-on-grade with sensitive flooring to prevent this moisture from entering the building above. 
In summary, agglomerated tiles offer an appealing appearance similar to natural stone at a lower cost. To achieve successful installations, manufacturers should provide design data for moisture-related dimensional changes, specifiers should require in-situ moisture testing similar to those used for other flooring materials, and the industry should develop standards for fabrication and installation of agglomerated tiles. BD+C
 
About the Authors
Justin S. deWolfe, P.E. (jdewolfe@sgh.com), is a Staff II engineer at national engineering firm Simpson Gumpertz & Heger Inc. (SGH). His experience includes building envelope and science investigations, designing and investigating repairs for historic structures, and floor slab investigations. 
Rene W. Luft, Sc.D., P.E. (rluft@sgh.com), is a Senior Principal at SGH. He specializes in investigations of building and material problems, exterior walls, moisture and mold, flooring failures, and post-tensioned concrete buildings and foundations
         
 

RELATED ARTICLES FROM BD+C

Comments on: "WEB EXCLUSIVE: Moisture-related failures in agglomerated floor tiles"