Reducing the costs of conflicting tolerances for building slabs and installing stairs

Every building taller than one story has stairs of some kind. Whether a dramatic entry feature or an exit requirement, stairs are a common component for designers with explicit requirements, yet stairs can become complicated and challenging to construct for many reasons.

Requirements and challenges Stair tolerances given in the National Fire Protection Association’s Life Safety Code are easy to understand. Section states that, “Riser height shall be measured as the vertical distance between tread nosings.” Illustrations in the Life Safety Code show how riser height is to be measured when the tread slopes to the front or the back. Section then states that, “There shall be no variation in excess of 3/16 inch in the depth of adjacent risers, and the tolerance between the largest and smallest riser shall not exceed 3/8 inch in any flight.” These apparently simple requirements, however, can cause major problems for contractors, including the following:

  • the requirements limit the tolerance that can be applied at the levels of the building connecting to the stairs;
  • elevation tolerances for cast-in-place, suspended concrete slabs do not match the riser tolerances;
  • no measurement protocol is given, therefore, no guidance is offered on how many measurements to take, or where the riser measurement should be located along the width of the stair; and
  • the wording is such that there is no allowance for out-of-tolerance situations.

Section can be interpreted to mean that if any riser measurement taken along the entire width of each step exceeds any riser measurement taken along the entire width of either of the adjacent steps, such variation is unacceptable.

The controlling tolerance problem

Section 4.4.1 of the soon-to-be published American Concrete Institute’s (ACI) Specification for Tolerances in Concrete Construction and Materials (ACI 117-06) states that the elevation tolerance for formed slabs, before removal of supporting shores, is ±3/4 inch. Section 4.7.1 of ACI 117-06 includes the same ±3/16-inch tolerance for the depth of consecutive risers as is used in the NFPA Life Safety Code.

Based on these permitted values, it’s possible for the elevations of consecutive floors to differ by 1-1/2 inches and still be within elevation tolerance, but it’s very unlikely that the ±3/16-inch tolerance requirement will be met, as illustrated in the example below.


Assume that the height from finish floor-to-finish floor is 120 inches and precast stairs with a riser height of 7-1/2 inches are used. If the floor-to-floor elevation is actually 121- 1/2 inches, the extreme limits of the floor elevation tolerances, the extra 1- 1/2 inches has to be accommodated at the top or bottom of the stairs.To guard a fall from the top of the stairs— most contractors will match the top of the stairs to the floor elevation there and shim the bottom (see Figure 1). That means that the riser for the first stair will be 9 inches high, far exceeding the allowable 3/16-inch allowable variation in the height of adjacent risers. If the floor-to-floor elevation difference goes the other way, and is only 118-1/2 inches, the fix will also have to be made at the bottom, perhaps by chipping out concrete. Then, the precast stairs could have a riser height at the bottom of only 6 inches. These are the extreme cases, but only when the two floor elevations are within 3/16 inch of each other will prefabricated or precast stairs fall within the allowable tolerances.

Are the ACI tolerances on floor elevation too lenient? Based on as-built data, Birkeland and Westhoff concluded in their 1971 article “Dimensional Tolerances in a Tall Building” that a realistic tolerance on deviation from intended elevation in a well-constructed building might be ±1 inch. In discussing this and other results of as-built measurements, the authors concluded, “We feel this building is a good representative of high-quality construction. The floors look flat and feel flat. There have been no undue problems with doors or partitions.

Hence, if you wish to criticize the workmanship of this building, please have ready comparably complete measurements on one of your buildings.” The current ACI tolerance on intended floor elevation is less than the ±1-inch tolerance suggested by Birkeland and Westhoff, and their other comments are still applicable. So I think that it is likely that today’s contractor doing high-quality construction will still have problems making stairs fit.

What are the contractor’s options when floor elevations differ? One is to use cast-in-place stairs and build the formwork to match the elevation differences between floors. This used to be the way contractors handled differences in planned and as-built dimensions, but it has two major disadvantages. Cast-inplace stairs take more time to install than manufactured or precast stairs, and are often more costly. They are seldom an option if there is a tight construction schedule. This is especially true if carpenters have to custom build forms to accommodate floor-to-floor elevation differences, spreading the differences over enough risers to meet the 3/16-inch and 3/8-inch tolerances.

It’s possible that several different sets of forms with slightly differing dimensions would be needed. On jobs where cast-in-place forms are used, carpenters are most likely to build a standard set of forms and simply shim or cut the forms to make them fit. Thus, the situation is no different from that for pre-manufactured stairs.

Another option is adjusting the elevation of the floor slab at the landing by adding a sloped topping.This may work well in enclosed stair wells. But if the stairs open into a hallway, such elevation adjustments are likely to be unsightly.

This corrective measure may also result in a feather edge for the topping unless substrate concrete is chipped out under the thin edge of the topping.

How will the riser height be measured? The illustrations in NFPA Life Safety Code show that the height of the riser is the vertical distance between tread nosings, and that the point of measurement differs for treads sloping to the front or back. For treads that slope to the front, the measurement is made at the back of the tread. For treads that slope to the back, the measurement is made from nosing to nosing—presumably with the aid of a carpenter’s level (see Figure 2). Most inspectors I’ve seen measuring floors just measure the riser height at the back of the tread.

The location of the measurement along the tread width isn’t specified, but it should be. With a tolerance of only 3/16 inch, it’s easily possible to use up a third or more of the tolerance because of variations in flatness on the tread. Because of the current specification threshold value of 3/16 inch, it seems sensible to at least have the measurements taken where the foot traffic is most likely. Requiring corrective action for an out-of-tolerance spot at the end of the tread can be an extreme response to an already extreme tolerance requirement.

One approach would be to just measure at the centerline of the tread width or at one or two points in the middle third of the tread, then average the measurements. An even better approach would be to make the tolerances realistically attainable.

Is the tight riser-height tolerance needed? Some stair safety consultants propose that, “It’s unthinkable to suggest that some code reforms should be made lenient on stair tolerances,” and ask, “Isn’t it simply more sensible and more important to build and maintain stairs that comply with strict, effective safety standards than to just save money and create unnecessary injury, pain, and suffering?”(Ubell, A. and Ubell, L., “Watch Your Step and Hold On,” Accurate Building Inspectors Newsletter, Vol. 2, No. 4, Spring 2005) I would argue that there are no studies showing that a difference of 3/16 inch in the height of adjacent risers is less likely to cause a trip-and-fall accident than a difference of 1/4 inch or more. So the effectiveness of the strict safety standards is unproven.

A 1992 NAHB Research Center review of literature and data for a stair safety study titled, “Stair Safety—A Review of the Literature and Data Concerning Stair Geometry and Other Characteristics,” that was prepared for the U.S. Department of Housing and Urban Development, yielded only five research efforts aimed directly at identifying the role of riser and tread geometry in residential and industrial stair safety through either the study of accidents or some proxy such as “incidents.” An indepth review of these major works found that the described research, considered either separately or together, failed to establish a consistent, statistically valid link between stair safety and stair geometry.

Without such a link between riser height and stair safety, it is hard to support an argument that the 3/16-inch tolerance on riser height is necessary.

In Birkeland and Westhoff ’s words: “Nobody seems to know what tolerances are realistically obtainable, nor what tolerances are actually required to obtain a satisfactory building.” I believe the latter part of this sentence is applicable to riser tolerances: Nobody knows what tolerances are actually required to make the stairs safe, so when codes allow no variation from the riser height tolerance, they create a building cost with no commensurate benefit. And pity the poor building inspector who has to order rework because one stair riser is 1/8 inch too high.

The inspector’s dilemma when determining compliance with the current building safety code, building inspectors can only measure a riser, compare that measurement with one on the adjacent riser, and then verify that the difference meets the code requirement.

If it doesn’t, rework must be ordered. Inspectors have no authority to say the measurements are close enough, because that isn’t their job. Their job is to ensure that the completed building meets code requirements.

If a statistical approach were followed, we could allow a low percentage of out-of-tolerance measurements.

This allowance is nothing more than an admission that nothing can be built with absolutely uniform dimensions, and—even when we set tolerances on dimension uniformity—an occasional variation that is outside that tolerance should be expected. We simply set limits on the magnitude and frequency of such variations. Strength requirements for concrete are examples of this approach. We design a building on the assumption of a specified compressive strength. When the strength is tested, the code requires the following:

  • the average of any three consecutive test results must equal or exceed the specified compressive strength, and
  • no individual test result must fall more than 500 pounds per square inch below the specified compressive strength.

This requirement results in a 90-percent probability that any individual test result will equal or exceed the specified compressive strength, and a 99-percent probability that the consecutive- test and minimum strength requirements will be met. Note that it makes inspectors’ jobs much easier.

They’re never required to reject completed concrete work because of a relatively small deviation from the specified strength.

A similar approach could be used for measuring dimensions of stairs. The difficulty would probably not be in decisions regarding the frequency or magnitude of deviations from dimensional tolerances. It’s more likely that emotional arguments would be somewhat along the lines of, “How can you even consider allowing these deviations to reduce construction costs and delays, when a trip-and-fall accident could result in serious injury or death?” A suggested approach Instead of trying to achieve an almost uniform riser height over the full set of stairs, a more reasonable and attainable approach is to allow a greater tolerance for the riser height of the bottom stair. Requiring no more than say a 3/8-inch or 1/2-inch deviation instead of the currently required 3/16- inch deviation from the height of the adjacent riser would have several advantages as follows:

  • it would affect only the bottom step, where a stair accident is less likely to cause serious injury;
  • it would not change the rhythm or gait of the stair users for their walk up the remainder of the stairs because stairs fabricated offsite can be built to more exacting tolerances, and the existing 3/16-inch tolerance could be retained; and
  • it would reduce the cost of rework and delays resulting from the rework.

Establishing a measuring protocol for riser height is also needed. And finally, a statistical approach to acceptance or rejection of as-built stair dimensions is desirable, but might be the most difficult to achieve.

Bruce Suprenant, Ph.D., P.E., is president of Concrete Engineering Specialists in Boulder, Colo. He can be reached at 303-499-0264 or