Hospital Fit-Out: Plenum, Conduit Sharing, and a Nine-Day Deadline
Tenant improvement on the third floor of a hospital: 48 new fiber-to-the-desk outlets, each a 55 m run of 6-strand OM4 from a telecommunications room to a patient-room media panel. The ceiling above the corridor is a return-air plenum and the corridor wall is a 1-hour rated assembly. The electrical subcontractor has installed 3/4 in. EMT containing four Cat 6A cables and proposes to pull the fiber into the same conduit, since fiber does not carry current. The installing contractor's plan calls for OFNR riser-rated indoor cable because it is cheaper and it is inside conduit anyway, steel fish-tape pulls through four 90-degree bends, nylon tie-wraps to ceiling grid wire every 1.5 m, and no firestop at the corridor penetration because the conduit is continuous. A 480 V feeder runs parallel 100 mm from the proposed pathway for 12 m. The customer's IT group wants all outlets labeled and tested before occupancy in nine days.
1. Rule on the cable rating and the shared conduit. What does the code allow, and what do you specify?
Read the worked answer
The ceiling is a return-air plenum, so NEC 300.22(C) governs anything installed in that environmental-air space. The code does permit metal raceway as a wiring method there, so OFNR inside a complete, continuous EMT run is arguable, but 300.22(C) also requires wiring to be listed for the purpose, and many authorities having jurisdiction read that as requiring plenum-listed cable regardless of the raceway. On a hospital project I would not build a design on an arguable reading: specify OFNP/OFCP for anything in that ceiling, and get the AHJ interpretation in writing before ordering if cost is contested. The shared conduit is a separate problem and is easier to reject. Four Cat 6A cables at roughly 7.5 mm OD need about 177 mm2 of cross-section, while 3/4 in. EMT has an internal area near 343 mm2, so the 40% fill allowance of about 137 mm2 is already exceeded before any fiber is added. Adding fiber makes an overfilled raceway worse, increases pulling tension on both cable types, and makes future adds impossible. Require a separate pathway, and keep a documented fill calculation in the submittal.
2. Assess the installation method: fish tape, four 90-degree bends, tie-wraps to grid wire, and the parallel 480 V feeder. What do you require before work starts?
Read the worked answer
Several items in that method would damage the fiber or fail inspection. A steel fish tape must never be attached directly to a fiber cable; use a pulling eye or pull sock with the tape or rope, feed from the far end with a second person, and keep tension below the manufacturer's limit. Four 90-degree bends in a short run create cumulative sidewall pressure; replace with a routed pathway using J-hooks or tray, and respect the bend radius. For a typical 6-strand tight-buffered cable around 5 mm OD, plan on roughly 50 mm radius installed and about 100 mm during the pull, verified against the datasheet. Do not support cable from ceiling grid wire with nylon tie-wraps: grid wire is not a listed support, the ties point-load the jacket, and seismic movement will abrade it. Support independently at roughly 1.0 to 1.5 m intervals per TIA-569 practice. The 480 V feeder 100 mm away is not a code violation for nonconductive fiber, but it is a practical and safety problem; maintain the facility's separation practice and route away from it where possible. Add firestopping at the rated wall penetration per the listed system, even if the conduit is continuous.
3. What does acceptance look like in nine days, and what happens if a strand fails?
Read the worked answer
Acceptance has three parts and none of them can be skipped because of the deadline. First, documentation and labeling per TIA-606: every outlet, panel port, and strand identified, with as-builts showing the actual pathway. Second, testing: 100% of installed strands, both directions where the method requires it, using an OLTS per TIA-526-14 practice, with results recorded against a stated loss budget rather than a pass/fail light. Include endface inspection of every connector with a scope, because a connector that passes loss can still be contaminated and fail in service. Third, firestop inspection and a walkthrough of supports and bend radii. If a strand fails, do not accept a re-test after simply re-plugging: inspect and clean the endface, re-test, and if it still fails, find the cause with an OTDR before replacing anything. Keep spare pre-terminated assemblies on site. Given nine days, sequence testing room by room as each is complete rather than batching at the end, so a failed strand is found while there is still time to fix it. Issue a punch list with photographs and a re-test record.