Sheet metal flashing is where commercial roofs leak: a field foreman primer

I have torn off more commercial roofs than I can count, and the leak is almost never where the building owner thinks it is. The membrane gets blamed because it is the big visible thing on the roof. The flashing causes the leak because nobody on the crew was ever walked through why the detail matters.

TL;DR: Most commercial roof leaks are not membrane failures. They are sheet metal flashing failures at parapets, penetrations, and terminations. SMACNA and NRCA publish detail sets that solve about ninety percent of the problem, and crews still install parapets wrong because nobody walks them through the why. A foreman primer that respects what the crew already knows beats another binder on the truck.

The membrane is not your problem

Pull the cover board off a leaking commercial roof and trace the water back. Nine times out of ten the trail starts at a vertical termination, a curb, a drain bowl, or a parapet cap. The field of the membrane is fine. The TPO weld passed the probe. The EPDM seam tape is still tight. What failed is the sheet metal.

This is not a controversial position inside the trade. NRCA has been saying it in print for thirty years. SMACNA puts the details in their Architectural Sheet Metal Manual and republishes them every few years. Every roof consultant report I have ever read on a problem building ends in the same place. The leaks live at the flashings.

And yet I still walk roofs where the parapet was wrapped by a guy who watched a YouTube short, the termination bar was shot in at twenty four inch centers, and the counterflashing was face fastened with no reglet. The membrane manufacturer takes the warranty hit. The crew gets blamed. The real problem is that the field foreman was never taught the why.

Why parapet flashing fails first

The parapet is the worst piece of real estate on a commercial roof. It moves more than anything else up there. Sun on one side, shade on the other, two different expansion rates fighting over one piece of metal. Wind pulls up on the cap. Snow drifts pack against the inside. Rain runs down the back side every time it pours.

Three things kill parapet flashing.

First, the membrane gets terminated too low. The top of the membrane needs to sit at least eight inches above the finished roof surface, and the cap flashing needs to extend down past that termination by at least four inches. I see crews land the termination at four inches because the parapet is short and nobody wanted to fight it. Water finds that joint inside one freeze cycle.

Second, the cap is not back sloped. A flat or front sloped coping pushes water onto the face of the building and back behind the termination bar. The NRCA detail calls for a quarter inch per foot slope toward the roof. On a twelve inch wide cap that is a quarter inch of drop. It takes thirty seconds to set the cleat right and crews still skip it.

Third, the joints in the coping are butted with sealant and called done. Sealant is not a joint. Sealant is a backup to a joint. A proper coping uses standing seam, double lock, or a backer plate with a cover plate. The cover plate moves with the metal. Sealant cracks the first hard winter, and now you have a funnel pointed straight down inside the wall cavity.

Termination bar spacing is not a suggestion

The spec sheet on every major single ply manufacturer says termination bar fasteners go on eight inch centers, sometimes six. I have seen crews shoot them at twelve and at sixteen. The bar looks tight when you stand on it. It is not tight in a fifty mile per hour gust.

The point of close spacing is not to hold the membrane on. The point is to keep the bar pressed flat against the substrate so the sealant bead behind it stays compressed. As soon as the bar bows out between fasteners, the bead breaks contact, and you have a capillary path. Water does not need a hole. It needs a gap and a pressure difference. The wind gives it both.

The fix is boring. Mark your spacing with a crayon before you shoot. Use the right anchor for the substrate. Run a fresh bead of urethane behind the bar, not the cheap stuff in the gun box from last week. And butt the bar sections tight with a one eighth inch gap and a dab of sealant at every joint. That is the whole job. It takes an extra ten minutes per hundred feet and it is the difference between a roof that lasts twenty years and one that calls you back in three.

Counterflashing on TPO vs EPDM

The counterflashing detail is not the same for thermoplastic and thermoset roofs and crews mix them up all the time.

On TPO, the base flashing is welded to a coated metal flashing or a TPO clad termination strip, then a separate counterflashing in painted steel or aluminum laps down over the top. The weld is the seal. The counterflashing is decorative and a wind shield. If your welder cooked the membrane brown at the base, you do not need the counterflashing to save you. If the weld is bad, no amount of sealant under the counterflashing will hide it for long.

On EPDM, the base flashing is adhered to the wall with bonding adhesive and the termination is mechanical, usually a termination bar with a lap sealant cover. The counterflashing here is part of the system. It physically holds the top edge of the EPDM in place. If you face fasten counterflashing into EPDM without a reglet or a proper receiver, you have just shot ten holes through your roof and called it a detail. Use a two piece counterflashing with a receiver set in a reglet cut, or a surface mounted receiver with a continuous cleat. The cover piece snaps into the receiver. No fasteners through the membrane.

I have watched crews try to detail an EPDM parapet like a TPO parapet and a TPO parapet like an EPDM parapet, and both buildings leaked inside a year. The systems are not interchangeable. Read the manufacturer detail, not the one in your head.

When soldered copper still beats sealant

There is a generation of foremen who have never soldered a joint and a generation of building owners who have never paid for it. Both groups assume sealant is the modern answer. It is not.

Sealant has a service life of seven to ten years on a good day. Soldered copper has a service life of fifty to one hundred years. On built up gravel stops, on through wall scuppers, on conductor heads, on chimney crickets above commercial kitchens, on any sheet metal that has to hold water and not just shed it, solder is still the right answer. SMACNA gives you the detail. The Copper Development Association publishes a free handbook that walks you through the joint geometry.

The places I will pay for soldered copper every time. Conductor heads on the front of an office building. Through wall scuppers on a parapet. Chimney saddles on a steep slope tie in. Decorative bay window pans on a historic restoration. Anything where the metal sits in standing water for any length of time.

Sealant is fine on a coping joint cover plate where the metal sheds water in seconds. Sealant is a disaster on a scupper throat where the metal holds water for hours during a storm. Know which one you are looking at.

How to actually train a crew on this

You cannot fix this by handing a guy a SMACNA manual. SMACNA is twelve hundred pages and it is written for architects and shop fabricators. Your foreman is not going to read it on his lunch break. NRCA details are better, but the binders are still where good intentions go to die.

Here is what works. Pull two crews off the truck. Walk a finished parapet on a building you are proud of. Walk a leaking parapet on a callback. Point at the differences with your finger. Six inches of difference in termination height. A back slope you can see with the eye. A standing seam cover plate versus a butted sealant joint. The crew will remember the visual for ten years. They will not remember the page number.

Then make a one page field card for the three details that fail the most. Parapet termination, penetration boot, drain bowl. Laminate it. Stick it in the cab of every truck. Make the foreman initial it on every job startup. That is the whole training program. It costs forty dollars and it saves you forty thousand a year in callbacks.

The practical takeaway

If you run a commercial roofing crew or you own a building that keeps springing leaks at the same spots, here is the short list.

1. Stop blaming the membrane until you have traced the water back to the field. Most of the time you will end up at a flashing.
2. Pull your last five leak callbacks. I bet four of them are parapet, penetration, or drain. That tells you where to spend your training money.
3. Match your detail to your system. TPO and EPDM use different counterflashing strategies. Do not mix them.
4. Pay for solder where solder belongs. Conductor heads, scuppers, chimney saddles. The math works over the life of the building.
5. Build a one page field card for your three worst details. Make every foreman sign off on it at job startup. The card beats the binder every time.

None of this is exotic. SMACNA published most of it before I was born. The reason the work goes sideways is that nobody on site was ever walked through why the detail exists. Fix the why and the how takes care of itself.

Bring this to your team

I have spent twenty plus years running commercial roofing and sheet metal crews, and the lessons that move the needle are the ones the field foreman can use on Monday morning. If you want me to come walk a problem building with your team, run a training session on flashing details that hold up in real weather, or speak to your operations group on what separates the crews that get called back from the ones that do not, book a conversation here.