Ballasted EPDM Re-Cover on a Joist-Framed Warehouse
An owner of an unheated 18,000 sq ft warehouse (steel joists at 5 ft o.c., 1.5-in. wide-rib steel deck, 32 ft tall) wants a ballasted EPDM re-cover instead of a fully adhered roof because he has heard adhesive roofs fail in cold weather. The parapet is only 8 in. above the existing roof surface. Design data: basic wind speed 120 mph, Exposure C, and the structural engineer has capped all added dead load at 14 psf. Your proposed assembly is 1.5-in. polyiso (about 0.8 psf), 60-mil loose-laid EPDM, #4 aggregate ballast at a nominal 10 psf in the field, and 18-psf concrete pavers across a 5-ft-wide perimeter band. The perimeter measures 700 lf, so the paver band is 3,500 sq ft and the field is 14,500 sq ft. The general contractor's estimator says the assembly is 'well under the limit' and wants to start Monday. Rain load and ponding are not included in the 14 psf figure.
1. Verify whether the proposed assembly satisfies the 14 psf added-load limit, and state whether the average load is the right test.
Read the worked answer
Run the numbers before arguing about anything else.
- Field: 14,500 sq ft x 10 psf = 145,000 lb
- Perimeter band: 3,500 sq ft x 18 psf = 63,000 lb
- Total ballast: 208,000 lb / 18,000 sq ft = 11.6 psf
- Add polyiso (~0.8) and membrane (~0.25): about 12.6 psf average
So the average passes, and the estimator is technically right and practically wrong. The engineer's cap is not a single global number applied to a whole building; each joist and each deck span carries its own tributary load. A perimeter joist under the 5-ft paver band sees roughly 18 + 0.8 + 0.25, or about 19 psf, which is 5 psf over the cap before you add rain or ponding. That is the number that matters.
Course of action: send the engineer the zone-by-zone loads, not the average, and get written confirmation for the perimeter and corners. If the perimeter cannot take 19 psf, either narrow the paver band, use lighter pavers, or take the perimeter to a fully adhered or mechanically attached edge and keep ballast only in the field. Do not let the crew mobilize on an average-load argument.
2. The 8-in. parapet: what performance problems does it create for a ballasted assembly, and what would you do about it?
Read the worked answer
Two separate problems converge at an 8-in. parapet.
First, flashing height. Base flashing normally extends at least 8 in. above the roof surface, and that dimension only works if the finished ballast surface is not sitting in the middle of it. With 1.5 to 2 in. of aggregate plus pavers, the freeboard above the finished surface drops to about 6 in. The installer is then forced into one of two bad outcomes: terminate the membrane below the required height, or bury the flashing and termination detail under ballast, where it cannot be inspected and where ballast movement can abrade it.
Second, ballast retention and scour. A low parapet does not reliably hold aggregate, and granular ballast blown off the perimeter both removes the uplift resistance exactly where demand is highest and becomes wind-borne debris at 120 mph Exposure C. Corners are the worst case.
Practical options: raise the perimeter with a wood nailer and metal fascia/gravel stop to gain height and a proper termination, and switch the perimeter and corners to pavers or an adhered/mechanically attached edge; or abandon ballast and go fully adhered or mechanically attached over the whole roof. Document the choice in writing to the owner.
3. The owner now asks to add a second layer of aggregate 'for wind safety' and to keep the pavers only at the corners. How do you respond?
Read the worked answer
A second layer is the wrong lever, and the reasoning is worth showing the owner.
Weight: a second 10 psf layer puts the field at roughly 21 psf plus insulation and membrane, about 22 psf against a 14 psf cap. That is a structural exceedance, not a detailing preference. It also buries the flashing deeper.
Performance: added ballast does raise uplift resistance, but it is an inefficient way to buy it, because the uplift demand it is answering is concentrated at corners and perimeters. Uniform extra weight in the field does almost nothing for the corner where the peel starts, while loading the structure everywhere.
Better answers, in order of preference: (1) put the money into the perimeter and corner zones with pavers or an adhered/mechanically attached edge and a proper perimeter securement detail, keeping field ballast at the approved 10 psf; (2) if the owner wants a single reliable system, change to fully adhered or mechanically attached and delete ballast; (3) if ballast must stay, get the manufacturer's ballast and perimeter enhancement requirements for 120 mph Exposure C and have the structural engineer approve the zone loads.
Separately, remind the owner that removing pavers from the perimeter to save weight and adding a field layer is exactly backwards for wind.