Rooftop Drain Pan and Condensate Routing

Purpose

On a rooftop packaged unit the condensate complaint is almost never a plugged line. It is a trap that is too shallow for the negative pressure the blower pulls at the pan, or a discharge dumped onto the membrane where it stains, ponds and eventually becomes the roofer's argument with the landlord. Both defects are invisible on a dry day and both were built in on install day. This procedure makes the tech measure the pressure the trap has to hold against, compare it to the trap that is actually there, and put the water somewhere the building owner has agreed to, so "drain checked" stops meaning "I looked at it."

Scope

Covers drain pans, traps and condensate routing on packaged rooftop units and rooftop air handlers on strip retail, small offices and restaurants: trap sizing against measured static, pan slope, roof-side termination and freeze exposure.

Does not cover the pour test method, float switch testing or condensate pumps in general, owned by the condensate drain service and overflow prevention SOP; this one cites it rather than re-deriving it. Does not cover roof drain or membrane repair, which is the roofer's work.

Roles and responsibilities

Procedure

1. Isolate the unit before any panel comes off. Open the unit disconnect, lock and tag it, and prove dead with the before-and-after instrument check per NFPA 70E-2021, 120.5, under 29 CFR 1910.333(b)(2). Acceptance: under 1 volt at the load-side terminals on a meter proved live on a known source before and after, since induced voltage on a de-energized conductor routinely reads a fraction of a volt. What wrong looks like: working a wet pan next to a live control compartment. Stop rule: no live reading afterward voids the test and you start over. Hazard: standing water in a pan sits directly under energized controls on most packaged units, so the isolation happens before a hand or a wet vacuum hose goes into that compartment.

2. Establish whether the pan sits on the suction or discharge side of the blower. A draw-through unit puts the coil and pan under negative pressure and needs a trap that can hold against it; a blow-through pan is positive and traps the other way. Acceptance: configuration named on the ticket, and the negative or positive static at the pan section measured in inches of water column with the panels closed and the blower running. What wrong looks like: assuming draw-through because most packaged units are. Stop rule: if there is no port to read through and the manual gives no figure, record that and use the manufacturer's rated external static as the working number, labelled as taken from the manual. Hazard: the reading is taken with the unit running, so the manometer goes on an existing port and no panel is opened to reach it.

3. Compute the trap the pan needs, then measure the trap that is there. For a draw-through pan the common form, and the one most manufacturers publish, is an inlet leg depth of at least the negative static in inches of water column plus 1 in, with the outlet leg at least half the inlet depth so the seal survives shutdown. The unit's installation manual governs where it gives a figure. Acceptance: computed depth written down beside the measured depth of the installed trap. What wrong looks like: a hardware-store P-trap holding about an inch, which the blower empties in seconds. Stop rule: a trap shallower than computed does not get cleaned and left; it gets rebuilt on this visit or the unit is written up. Hazard: measure the pressure with filters as found, since a loaded filter drives the negative deeper, and size the trap for that worst case rather than the clean-filter number.

4. Check that the unit and pan actually drain toward the outlet. Level the unit on its curb the way the manufacturer specifies, usually level or pitched slightly toward the drain connection. Acceptance: no standing water deeper than about 1/8 in anywhere in the pan ten minutes after the blower stops. What wrong looks like: water sitting at the end away from the outlet, which is where rust starts and where growth establishes. Stop rule: a curb out of level enough to pond gets shimmed or written up as a rigging correction, not compensated for by drilling a second drain. Hazard: reaching across a coil to feel for water puts a forearm on fin edges, so use a mirror and a light rather than a hand.

5. Clear the pan and prove flow with a measured pour. Run the pour test the condensate SOP specifies and record the volume and the time to appear at the termination. Acceptance: the poured volume appears at the discharge point within the time that SOP sets, with no rise in the pan. What wrong looks like: a line that takes the pour but slows, which is a partial occlusion that drains a trickle fine and backs up on the first hot afternoon. Stop rule: no flow at the termination stops the visit at the drain, not at the coil. Hazard: scrubbing, vacuuming or rinsing a slimed pan drives that biofilm airborne, so wet the surface first, use a HEPA-filtered vacuum rather than compressed air, and where growth is heavy enough to shed on contact the work needs respiratory protection under a written program per 29 CFR 1910.134, which also means the crew does not do that work without one.

6. Follow the discharge to where it actually lands and get it approved. Walk the line to its end and look at the roof under it. The International Mechanical Code, in the edition your jurisdiction has adopted, requires condensate to discharge to an approved location and not where it creates a nuisance; on a roof, approved is the authority's call and in practice means a roof drain, scupper or downspout rather than the membrane. Acceptance: termination photographed, distance to the nearest roof drain measured, and no discharge onto a walking path, an entry below, or a ponding area. What wrong looks like: a pipe stub pointed at the membrane three feet from the curb with a stain and algae under it. Stop rule: relocating a line across someone's roof needs the owner's written approval and supports that do not penetrate the membrane, so it is quoted, not improvised. Hazard: the stained area under a long-running discharge is slick, so route your footing around it rather than stepping over the pipe.

7. Judge the freeze exposure before winter, not during it. An exposed rooftop line that holds water freezes, splits, and leaks into the pan section in spring. Acceptance: the run either drains fully by gravity with no sag holding water, or carries exterior-rated heat trace proven to draw current. What wrong looks like: a long horizontal run with a belly, insulated but not traced, which is a full pipe wrapped in a blanket. Stop rule: a line that cannot be pitched to drain gets traced or rerouted, and the recommendation goes to the customer in writing before the first freeze. Hazard: heat trace is a branch circuit on a wet roof, so confirm ground-fault protection at its supply before you meter it.

8. Restore, run the unit and prove the overflow control still stops it. Panels back with every fastener, hands clear, lock and tag removed, disconnect closed from the hinge side after calling it out. Run in cooling at least 15 minutes. Acceptance: water at the termination, no water anywhere but the pan, and the float or overflow switch proven to shut the unit down when lifted, then reset with the unit restarting. What wrong looks like: leaving on the assumption the switch works because you did not disturb it, when you just changed the water level it sees. Stop rule: a switch that does not interrupt keeps the unit off until it is replaced, because the whole reason a rooftop overflow reaches a tenant ceiling is that nothing shut the unit down. Hazard: this is where energy goes back in with the pan open to view, so nobody is in the cabinet when the handle moves, and the switch is lifted with an insulated tool rather than a wet hand.

The record this produces

One condensate block per unit: draw-through or blow-through, measured static at the pan in inches of water column with the filter condition it was measured at, computed trap depth, measured trap inlet and outlet depths, pan standing-water check, pour volume and time to appear, termination type and distance to the nearest roof drain, a photo of the termination and the roof under it, freeze provision, and the overflow switch result. Anything not corrected on the visit gets a dated line and the name of who was told.

The termination photo earns its keep. Roof staining is a slow argument that surfaces two years later between the landlord, the roofer and whoever last touched the unit, and a dated photo showing the stain already present, or absent, settles it in a minute. The static and trap numbers do the other job: next season's tech compares the measured negative against this one, and a deeper reading at the same filter condition means the coil is loading.

Worked pass: 5-ton packaged unit over a nail salon, water reported at a ceiling tile

Ticket said a stained ceiling tile inside the tenant space, no water yet on the floor. Draw-through packaged unit, filters visibly loaded, last serviced by another shop.

Step 1 passed: disconnect open, locked, tagged, meter proved live at the curb receptacle circuit, 0.3 V at the load lugs, meter proved live again. Step 2 confirmed draw-through and read the pan section at 1.2 in w.c. negative with the filters as found and the blower running.

Step 3 FAILED. Required inlet leg is the measured 1.2 in w.c. plus 1 in, which is 2.2 in, rounded up to 2.5 in for build. The installed trap measured 1.5 in of inlet depth, which is 0.7 in short of the 2.2 in requirement. Under the stop rule the pan was not simply cleaned and closed. The trap was rebuilt in PVC at 2.5 in of inlet leg with a 1.25 in outlet leg, which is half the inlet depth as the step requires, and fitted with a cleanout plug. That single defect explains the complaint: at 1.5 in of seal against 1.2 in w.c. of pull, the blower drew the seal down and pulled air up the drain, the pan held water instead of draining, and it spilled over the low corner into the supply section and down the curb.

Step 4: pan checked ten minutes after shutdown, a film only, no measurable standing water, unit level within the manufacturer's tolerance. Step 5: one gallon poured, at the termination in about 20 seconds, no rise in the pan. Step 6: the line ended in a stub 3 ft from the curb pointed at the membrane, stain and algae under it, nearest roof drain paced at 22 ft. Photographed, measured and quoted as a relocation with supports rather than run that day, because the approval was not in hand.

Step 7: the existing stub drains fully by gravity, no belly, so no trace for this run; the note flags that the 22 ft relocation, if approved, must be pitched or traced. Step 8: panels back, disconnect closed from the hinge side with the roof called clear, 20 minutes of cooling, water running steadily at the stub, pan clear, float switch lifted with a fibreglass rod and the unit shut down within a second, then reset and restarted.

What the failure teaches: a tech who ran only steps 4, 5 and 8 would have found a clean pan, a drain that takes a gallon in 20 seconds and a working float switch, and would have left calling the drain good. Everything downstream of the trap was fine. The unit was overflowing because of a dimension nobody had ever measured against a pressure nobody had ever read.

References

• International Mechanical Code Section 307, condensate disposal, in the edition your jurisdiction has adopted; what counts as an approved rooftop location is the authority having jurisdiction's call
• The unit manufacturer's installation manual for trap dimensions and leveling tolerance, which governs over any rule of thumb
• 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5, for isolation and the before-and-after instrument check; 29 CFR 1910.134 for the written respiratory program any respirator use in a slimed pan depends on
• See related: Condensate Drain Service and Overflow Prevention; Light Commercial Rooftop Unit Service