The Air Handler That Cracked the Ceiling

Why this matters

Ceiling cracks under attic equipment get blamed on the house, the drywall finisher or the weather, and the customer gets a patch that cracks again. In this case three ordinary explanations were each tested and ruled out on evidence before the real one was found: the replacement air handler was sitting on truss bottom chords that were never designed to carry it. The order of elimination matters, because each test took minutes, and skipping to "it's the weight" without them would have left the shop unable to say why the other explanations were wrong when the customer's drywall contractor said otherwise.

The signal

Eighteen months after the shop replaced an attic air handler, the customer called about cracks in the hallway ceiling directly below it. Two straight cracks ran the width of the hall about 24 in apart, and three nail pops had appeared along each crack. The customer had the hall patched once, six months earlier; the cracks came back in the same lines.

The house: single story, roof framed with prefabricated wood trusses at 24 in on center, spanning 26 ft from exterior wall to exterior wall, with no interior bearing wall under them. The hall ceiling is 1/2 in drywall fastened to the truss bottom chords.

The old unit, a smaller horizontal air handler, had sat on the same spot for years. The replacement was larger, a horizontal air handler with an attached coil. The tech who installed it had built a new platform for it: a 30 in wide frame of 2x4 with a 3/4 in plywood deck, laid across the bottom chords of two trusses, with a secondary drain pan under the unit. Nobody had measured anything about the trusses.

Before entering the attic, the tech switched the system off at the thermostat and opened the air handler's disconnect, because the inspection would put hands near the unit and its condensate line.

Hypothesis 1: the house is settling

Settlement of the foundation moves a whole wall or corner, and it shows in the walls first: diagonal cracks running from the corners of doors and windows, doors that rub or will not latch, gaps at the top of trim.

Test. Walk the house. Check each door for swing and latch, look at the corners of every window and door opening, sight along the hall walls for cracks.

Result. All interior doors swung and latched. No diagonal cracks at any opening. No cracks in any wall. The only cracks in the house were the two in the hall ceiling, and they ran parallel to the trusses, not diagonally.

Verdict. Ruled out. A foundation movement that cracks only one ceiling, in two parallel straight lines, and nothing else, is not what settlement looks like.

Hypothesis 2: moisture is softening the drywall

Wet drywall sags and cracks, and an air handler in an attic is a moisture source: condensate, a sweating duct or a pan that has overflowed.

Test. In the attic, look at the secondary pan, the insulation around the platform, the duct connections and the roof deck above. Below, compare a pin-free moisture meter reading on the cracked ceiling against a reference spot on the same ceiling 10 ft away.

Result. The secondary pan was dry and clean, with no rust ring. Insulation around the platform was dry and evenly lofted. No stains on the ceiling side. The meter read the same on the cracked area and the reference spot, within one point on its relative scale.

Verdict. Ruled out. A moisture cause leaves a mark somewhere, and there was none.

Hypothesis 3: vibration from the unit

The installer had set the unit directly on the plywood with no isolation pads. A blower out of balance, or even a normal one, shakes the platform, and vibration loosens drywall fasteners and works cracks wider.

Test. Measure the crack widths with a crack-width gauge card at three marked points, run the blower for 30 minutes, measure again. Then look at where the nail pops were.

Result. The crack widths read the same before and after, at the gauge's finest step. And the nail pops were along the cracks only, not scattered through the hall, which vibration would tend to do across the whole area the platform shakes.

Verdict. Not the cause, though the missing isolation pads were a real defect and went on the fix list. Vibration can widen a crack that already exists, but it does not explain two straight lines 24 in apart, exactly the spacing of the two trusses under the platform.

Hypothesis 4: the chords are deflecting under the load

That spacing was the clue. Two cracks, 24 in apart, one under each truss carrying the platform, meant those two trusses had moved relative to their neighbors, and the drywall bridging between moving and non-moving chords cracked along the moving ones.

Test. In the attic, walking only on chords and the platform, the tech ran a string along the top of the bottom chord of each truss from bearing to bearing, on equal 1 in blocks at each wall, and measured string to chord at the platform location:

The two trusses carrying the platform sat 1/2 and 9/16 in low against 1/8 in for each neighbor, a difference of 3/8 to 7/16 in over 24 in of drywall. That is the crack.

The design load. Each truss carried a manufacturer's label. Through the label, the office reached the truss manufacturer, which sent the truss design drawing. Its loading block read a bottom chord dead load of 10 psf and a bottom chord live load of 10 psf, the second marked non-concurrent with other live loads (the usual basis for an attic without storage). At 24 in spacing, 10 psf of dead load is 20 lb per linear foot of chord. That allowance already carries the ceiling drywall and the insulation.

The added load. The spec sheet gave the air handler and coil at 120 lb. The platform weighed 40 lb on the shop's bathroom scale before installation, according to the original job notes; call it 160 lb in all, the secondary pan empty. Split across two trusses, that is 80 lb on each, spread over the 30 in platform width, or about 32 lb per linear foot of chord. So over that 30 in, each chord carried the equipment's 32 lb per foot on top of a design dead load allowance of 20 lb per foot that the ceiling already used: the equipment alone was about 1.6 times the chord's whole design dead load over that stretch. The truss designer's reply put it more simply: the drawing included no equipment load, and the added load exceeded what the bottom chord was designed for at that location.

The IRC (R802.10 in the 2021 edition, as adopted by your jurisdiction) treats this as a truss alteration: added load beyond the truss design is not permitted without verification that the truss can support it, and truss members are not to be cut or altered without approval of a registered design professional. The old, lighter unit had been inside whatever margin the trusses carried. The new one was not.

Verdict. Confirmed, on three independent pieces of evidence: crack spacing matching truss spacing, measured sag on exactly the loaded trusses, and a design drawing with no allowance for the load.

The fix the truss designer issued

The shop ran it through its structural approval procedure (see related: The Structural Cut Approval SOP). The truss designer's written detail:

1. Remove the platform and rebuild it to spread the unit's load across five trusses, bearing on the bottom chords at their panel points (the places where web members meet the chord), with the member sizes and fastener schedule the detail gave.
2. Set the unit on isolation pads on the new platform.
3. Do not jack the sagged chords. The detail called for no lift, because the chords were not damaged, and pushing a loaded chord upward risks the gusset plate connections.

The work. The unit, coil and pan were lifted off the platform by two people, never one, because 120 lb on attic chords is a fall and a sprain at once. Refrigerant lines stayed connected and were supported so they did not kink as the unit moved a few inches; the condensate line was disconnected and capped. Both people stayed on chords and on the old platform, and kept clear of the cable runs along the chords. The attic was over 100 F by early afternoon, so the work ran in the morning with water on hand and breaks out of the attic.

Putting it back in service. With the unit on its pads, the condensate and secondary drain lines reconnected, and the disconnect closed, the tech poured water into the secondary pan until its float switch opened and confirmed the system shut off at the thermostat call, then drained the pan and confirmed the system restarted. The float switch is the device that keeps an overflow from wetting the ceiling, and it had been moved with the pan.

How they confirmed it

The shop marked the string-line readings and returned after six weeks. The two chords read the same as at the fix, 1/2 and 9/16 in, which is what the designer expected: the detail stopped further movement rather than reversing it. The crack widths on the gauge card were unchanged over the same six weeks. Only then was the ceiling repaired, by cutting out the cracked tape lines, refastening the board to the chords with screws, and retaping. Patching before the six-week check would have tested nothing.

What changed in the shop after this job: any replacement of attic equipment heavier than the unit it replaces gets the truss label read and photographed at the proposal visit, and the old and new installed weights from the two spec sheets written on the proposal. If the new one is heavier and there is no truss drawing, the office requests one before the install date. See related: Hanging Equipment From Framing, which owns the load path, and Roof Trusses and Why You Never Cut a Web, which owns how a truss carries load.

References

• International Residential Code, 2021 edition, R802.10 (wood trusses, including truss design drawings and alterations), as adopted by your jurisdiction
• ANSI/TPI 1, National Design Standard for Metal Plate Connected Wood Truss Construction, in the edition referenced by your adopted code, which governs truss design and reaches the installer through the truss design drawing
• See related: Roof Trusses and Why You Never Cut a Web; Hanging Equipment From Framing; The Structural Cut Approval SOP; How to Move Through an Attic Without Stepping Through a Ceiling