Industrial floors work better if the joints in them have the ability to transfer vertical loads. Joints that transfer loads hold up longer under traffic, and they need less maintenance. In some cases, they allow the slab to support heavier loads.

People have known all that for decades, but the issue has received more attention in recent years. One big reason for that extra attention comes from the materials handling industry, which has been switching to vehicles with small, hard wheels that are tough on joints. Another reason is the practice, once rare but now common, of building unreinforced slabs that rely solely on aggregate interlock to transfer loads.

Thanks to this recent focus on load transfer, we now have a performance specification for concrete floor joints. According to the latest version of ACI 360, Guide to Design of Slabs-on-Ground, joints in industrial floors should be designed and built so that differential vertical movement, under normal service loads, stays below a specified value. The specified value is 0.010 inch where vehicles have small, hard wheels — which are the norm in modern warehouses. The value rises to 0.020 inch where large, cushioned wheels are used.

Meeting the specification

How do you make sure floor joints meet those new recommendations? If you are designing a new floor, you can choose from a variety of load-transfer methods and devices. You can rely on keyways and aggregate interlock, which are cheap but often fail to meet the limits recommended in ACI 360. You can specify steel dowels — either the traditional round-section bars, or square-section bars, or flat plates. Dowels generally satisfy the ACI recommendations, though their success rate is less than 100 percent. Less common options include tie bars and post-tensioning tendons.

But all of those methods and devices are meant for new construction. They don’t work in existing floors. What’s the answer when joints and cracks in an existing floor fail to transfer the load?

It’s a common problem, but before 2005 the solutions all had drawbacks. You could pack the joint or crack full-depth with semi-rigid epoxy. That often worked at first, but the problem came back sooner or later. You could try dowel-bar retrofit, a method borrowed from highway engineers. This involves cutting slots in the slab, placing dowels in the slots, and covering the dowels with epoxy mortar. This worked in every concrete floor where it was tried. But it’s a tricky job and never caught on outside highway restoration.

Another method sometimes offered as a solution to unstable joints was to pump grout under the slab. Strictly speaking, this was not a remedy for poor load transfer since it did nothing to establish a connection across the joint or crack. It aimed instead to reduce differential movement by shoring up the slab edges, a fundamentally different approach. Sub-slab grouting reduced differential movement at joints, but, as with semi-rigid filler, the improvement did not always last. All of the available methods — joint filler, dowel-bar retrofit, and sub-slab grouting — relied on site-mixed chemicals that took a while to harden. Floors had to be taken out of service during the repairs, sometimes for a day or more.

A third solution appeared in 2005. Called a joint stabilizer, this device worked purely by mechanical action, with no need for site-mixed chemicals. The reliance on mechanical action meant that stabilizers could be installed at any temperature, even in freezers, and that floors could go back into service immediately.

Joint stabilizer performance

Joint stabilizers worked. They reliably reduced differential movement (see sidebar) to less than 0.010 inch. But would they last?

Skeptics predicted that corrosion would lock up the stabilizers, that repeated loading would break the stabilizers loose from the surrounding concrete, and that joints would widen beyond the stabilizers’ movement capacity. But the inventors and sellers of the device were confident it would keep working for years. Who was right?

To help answer that question, I recently visited three buildings in Pennsylvania, New York, and Indiana where joint stabilizers had been installed. My list included the oldest installation, the first big installation, and a floor that was just a few weeks old when it got joint stabilizers. While visiting each site, I measured differential movement and looked for signs of loose stabilizers, damage to the stabilizers, and damage to the floor.


A steel fiber factory in western Pennsylvania contains the oldest joint stabilizers in use anywhere. This was a small project, consisting of just five stabilizers. Despite the low number of devices, I felt it was worth a look, mainly for its age, but also because the building is neither heated nor air conditioned. That means the floor undergoes wide swings in temperature.

The floor was built in 2001. It has a design thickness of 8 inches, and is reinforced with 1-inch steel fibers at a dosage of 100 pounds per cubic yard. The joints, all sawcut, are spaced 25 feet apart and do not contain dowels.

The very first stabilizers were installed here in the summer of 2004, but those were prototypes that soon failed. The next set, almost identical to the devices being sold today, were placed in January 2005. After that they sat untouched, except by forklift wheels, for eight years until my visit in February 2013.

The stabilizers still looked intact, and were firmly embedded in the floor. Figure 1 shows the movement readings. Before repair, the mean movement was 0.038 inch. That exceeds the current ACI 360 recommendation of less than 0.020 inch. (The forklifts here have big, cushioned wheels, so ACI 360’s higher limit applies.) The installation of joint stabilizers took the movement down to 0.001 inch. Eight years later, it had risen to 0.002 inch.

Clearly, the stabilizers are still working after eight years. The difference between 0.001 inch movement at installation and 0.002 inch eight years later is trivial, and could even be due to testing error. I hesitate, however, to describe this project as an unqualified success. At two of the five stabilizers, small spalls have appeared in the surrounding concrete. Similar spalls have been reported on other projects, though not on the other two described in this article. The spalls do not seem to affect the stabilizer’s performance, nor do they interfere with the owner’s operations. Still, it is easy to imagine that some floor owners would object to them. Several explanations for the spalling have been offered, including damage caused by core-drilling, stabilizers that are too long and stick out the bottom of the slab, and latent delamination under the troweled concrete surface. All remain unproven.