Observations: In recent years many job inspections have shown different types of cracks in the tile assembly installed with TCA F144-01. These cracks are traced to the plywood or sub-floor panels installed on both conventional and EWS framing systems. Installation of the CBU or Fiber Cement panels have been correct in all cases. In two cases, specifically, an uncoupling system was installed with the same results. Installation of the tile conforms to ANSI A108.5-99. The puzzling angle to these failures have two common themes, not necessarily related to each other.

Conventional Framing: In every case observed with conventional framing, the free span of the floor joist system was placed at or near its maximum to provide L/360 Live Load, and a total uniform dead load of 10 psf. Where lesser quality material was used, such as spruce pine fir, yet still engineered for a greater product such as Doug Fir, most of the problems have occurred. For example, Doug Fir #2, 2x10, the maximum joist span allowed achieving L/360 and 10 psf of uniform dead load is 15’ 5”. Spruce-Pine-Fir #2, the same design allows for 14’3”. Considering the correct dead load designs for ceramic tile, tables provided for 27 psf should be used. The Doug Fir 2x10 now is allowed to span 13’4”, and the Spruce-Pine-Fir at 12’4”. This leaves a large margin of error in the design of the joist system. Potentially a difference of 37” overall in difference between what is designed into the joist system and what actually is installed. You can see that architects and or engineers are considering the heavy loads for tile, yet the builder is economizing with materials without the understanding of what the results will be.

This is very common and leads to excess deflection and improper loading with ceramic tile when installed over a CBU or mortar bed. Adding insult to injury, in kitchens, often there is an island cabinet with a stone or tile top creating a concentrated load pointed on two or three joists. This creates a lower area in the center of the span bay area. With each of the conventional framing tile failures, conventional sub-floor plywood products were used. In some cases, alternate sub-floor products such as particleboard and OSB were found with the same results.

Engineered Wood Systems (EWS): EWS framing is fast becoming the choice of architects and engineers for residential and light commercial applications. Some of the lessons learned on the past few years have been related to hard surface flooring and ceramic tile installations. The use of EWS allows the architect to create large open areas in demand with today’s lifestyle. In some cases, joist systems can be allowed to span 30’ or so. The lack of knowledge when designing these systems can be devastating to the end user. Proper selection of the joist type and depth is critical. Most residential applications will call for a series 15 or 25 joist with a depth of 9 ½”. This is to replace a conventional member of 2x10. When looking at span tables for maximum allowable spans for floor joists, EWS framing with a performance rating of PRI 40 (L/480 live load, recommended by APA) is 16’6” placed at 16”oc, and 14’3” with conventional framing using Spruce-Pine-Fir #2, and 15’5” using Doug Fir #2. (Conventional framing uses L/360 live load deflection). Unfortunately, most builders and engineers do NOT research all of the recommendations of the joist manufacturer of the APA. These dimensions noted are for 10 psf of dead or permanent load. Remember, a tile assembly weighs approximately 9 psf alone (TCA F144-01), not considering the weight of the sub-floor material, mechanical and plumbing materials hung from below, or cabinets and furniture that may be placed within the same joist bay area. Where mortar bed installations are considered and installed, the mortar bed alone can weigh as much as 11 psf. Most installations of EWS are spaced at 19.2” oc as a matter of economics.

When reviewing the allowable spans from the same tables as noted above, the limit is 15’7”. Looking at dead or uniform loads one more time, EWS provides several different tables related to performance. Simple span and multiple spans, each table dedicated to a uniform load of 10 psf, and a table dedicated to 20 psf of dead load. Using multiple spans with the same joist, (supported at either end and placed over a carrying beam) can add some advantages. Comparing the 16’6” simple span limit noted above, 16”oc, with a multiple span, the total free span from carrying beam to sill plate is now 17’2”. This is accomplished by using the joist’s integrity over-all to create a stronger system. Looking at the tables requiring 20 psf of dead load, the same joist is allowed a span of 15’8”, 16” oc.

There is one theory that disputes this when one portion of the multiple-span is loaded substantially different than the other. By loading one span of the multiple span joist with a ceramic tile installation could result in a net difference of 10 psf or more on one side of the carrying beam. This leads to the theory of cantilever deflection and twisting of the joists. It is best to isolate the bay areas by employing simple spans to the areas to receive tile. Other options consider the upgrade to a larger joist member, either a more substantial series of joist or a taller web. These situations have been noted in the failures inspected; yet there is no conclusive evidence at this time pointing to these theories as reason for failure. The APA/EWS has voluntarily raised the minimum deflection of its systems to L/480, 33% stiffer that conventional framing requirements for ceramic tile. The MIA has established L/720 for all natural stone installations and the Canadian Tile and Marble groups have established L/720 for all ceramic and stone tile installations.

Deflection remains an undocumented problem in the industry. Common misuse and misunderstanding of span tables and the related requirements for ceramic tile is all too common. Although L/360 has been the standard for nearly 100 years, the only evidence is data from ASTM C627-93 Robinson-type floor tester. This is for a concentrated load and not the overall reaction of a complete floor system. Common sense should tell us that a floor joist system designed at L/360 and has a free span of 20’ will allow for as much as .67” of vertical movement in the floor. The performance of a tile floor assembly with these dimensions will be unacceptable to the consumer.

Sub-Floor Materials: “Standard plywood” (5/8” and ¾”) seems to provide the best of all sub-floor materials. Tongue and groove material must be installed in accordance with building codes and manufacturer’s instructions. This includes proper adhesive between the sub-floor panel and the joist. When the gaps in the sub-floor (normally provided by the design of the tongue and groove) are installed improperly, there is no room for expansion and contraction of the plywood. The same holds true of OSB, only the expansion and contraction is at a greater rate. At this time, it has become a point of controversy. Who becomes responsible for assuring the gaps in the sub-floor are proper and clean for installation of tile. Where the gaps were not proper, (1/8”) and TCA F144-01 methods are applied, there is insufficient room for expansion and contraction. Where the gaps were proper, either from work provided by others, or the tile contractor carefully placed the cuts, thin set mortar was allowed into the gaps rendering them useless. This has created some of the failures; visible cracks transmitting through the entire tile assembly directly over the sub-floor seams. Recent arguments suggest that a flashing of some type or even duct tape be placed over the gaps to prevent thin set mortar from filling them in. Installations where flashings or tape was installed, no cracks have appeared. (Approximately three years where other failures occurred as soon as 6 months.) The installations where OSB sub-floor panels were used with this method have had some success; however other situations concerning movement have been noted.

Conclusion: Although engineering for today’s construction has become more of a mystery to the tile industry. It is essential that the tile contractor understand at least the basics of framing and its requirements. The study of EWS is essential also. Understanding the different combinations of joists and sizes related to performance, along with the proper handling and installation seems to be a common misunderstanding in the building industry. With conventional framing, and education in framing basics including species of wood, could be the difference between success and failure of the most basic of tile installations. Learning to recognize sub-floor problems up front and treating them accordingly using up to date materials and methods provided to the industry.

Opinion: Unfortunately, the finish trades bear the brunt of ignorance in most cases. The manufacturers of tile and setting materials are having to finance replacement of tile where it is not warranted. A serious effort to study the effects of deflection, uniform loads, and different framing systems in their completed environment is critical to the long term success of residential and light commercial tile installations where wood framing is concerned. This will take a substantial effort by many to accomplish, yet worthy of the attention in the ongoing efforts to grow the industry at a successful rate.

We would like to thank David deBear for his fine work in the CTC program.

Reference Materials:

I-Joists for Residential Floors
APA Engineered Wood Association
Tacoma Washington

Western Wood Products Association
Western Lumber Span Tables
Western Lumber Product Use Manual
Portland, OR

ANSI For the Installation of Ceramic Tile 1999

TCA Handbook for Ceramic Tile Installation 2001

Actual Job Inspections of Ceramic Tile Failures 1998 - 2001