The collapsing pergola

Apr 2, 2007 by

The insureds recently had a pergola built in their back yard. It was a freestanding unit between the pool and the house.   Supported by six tapered round columns, 14 inches at the base and eight feet high, with sweeping beams and graceful arches, it was a dramatic statement. Less than six months later, the columns were splitting and cracking.

The columns were made of a spun-cast, glass-fiber reinforced polymer resin (FRP), manufactured with ground stone to resemble cast concrete.According to the manufacturer, this material is, pound for pound, stronger than aluminum, concrete, or steel.  The wall thickness at the top was ¾ inch and the wall thickness at the base was 1 ¼ inches.   According to the manufacturers calculations, a single column of these specifications had a load carrying capacity of 20,000 lbs.   A rough guesstimate of the total weight of all of the 4 x 6 and 2 x 6 beams supported was considerably less than half that.

Here is the root cause of the problem.  The town’s building code did not accept this type of column without supporting calculations provided by a licensed engineer.   Either the homeowner or his contractor hired a local engineering company to do the calculations.

Rather than do their homework (which a two minute Google search, or a five minute phone call to the column manufacturer would have satisfied), they took the easy way out and, like most of us, did what was comfortable and familiar to them.   This FRP stuff is new, but they knew the properties of concrete.   Therefore they calculated the strength of the columns, based upon filling them with concrete. Similarly, plan check was familiar with concrete columns and, since the engineer calculated the column capacities, they approved the plans for building the pergola.

The contractor then had his marching orders to fill the columns with concrete.  Normally a contractor would bring in a concrete pump along side the concrete truck and by using a grout hose, could have filled the columns in less than an hour.   For reasons that only he knows, the contractor instead elected to fill the columns by hand.  Mixing up one bag of Kwikcrete at a time in a wheelbarrow, and then carrying the concrete in a bucket up a ladder to dump it into the top of the columns, the contractor filled the columns.   Each bag of concrete mix was then tamped with a piece of 2 x 4 lumber around the central stainless steel securing rod.   This method of filling left voids within the concrete and lead to separation of the fine material from within the concrete matrix.   The local building inspector did not observe the filling process.

We examined the details of the Tuscan capitals and bases and found them to be well sealed against water ingress.   Freezing water between the column shaft and the concrete core was ruled out.   Additionally, there had been no significant earthquake activity during that period.   Local wind speeds and gusts were recorded as less than 50 mph for the entire period; the design was for a basic wind speed of 70 mph.

The coefficient of thermal expansion for concrete is on the order of 60 times ten to the negative 6th power inches per inch-degree Fahrenheit, while that for FRP is on the order of 6 to 12 times ten to the negative 6th power.  This is a five to ten-fold difference, depending on the direction of the fibers and concrete selection.   This expansion rate difference will result in the concrete and FRP expanding to different overall dimensions (length and circumference), in response to temperature changes.   The greatest day and night temperature swing was 53 degrees Fahrenheit, a condition that guaranteed cracks and splits.

This was the reason the manufacturer’s installation instructions expressly forbid filling the void.   “It is not permissible at any time to fill the interior of the column shaft with sand, concrete or any other material.”

All involved ignored this fact: the insured, the engineer, the building department, and the contractor.

The concrete and FRP expanded and contracted at different rates which cracked and split the columns.