1920s downtown office: PoE lighting with occupancy analytics

The building

This is a representative scenario built from a type of building found on downtown streets across North Mississippi. The numbers are design-stage projections, not measured results.

The building is a two-story brick office from the 1920s, about 24,000 sq ft, with tall windows, plaster ceilings and a basement. Its owner-occupant, a regional professional-services firm, is renovating it for its own staff and weighing whether to lease part of the second floor. The current lighting is fluorescent from a 1990s fit-out, on wall switches with no sensors. Connected lighting load is about 28 kW, and the lighting uses roughly 80 MWh a year.

The firm’s IT manager wants the lighting on the company network, managed by IT, and wants to know how the building is actually used before anyone signs a sublease.

Constraints

• Preservation. If the renovation uses historic tax credits, visible changes to ceilings, windows and finishes go through preservation review. Surface conduit across a plaster ceiling won’t pass.
• Little ceiling space. No drop ceiling in the main rooms, shallow chases, and a few original pressed-metal ceilings.
• Old materials. Plaster and paint of this age are tested for asbestos and lead before anyone drills.
• Short runs. One network closet per floor puts every fixture within the 100 m Ethernet channel limit. That is the main reason PoE fits here, and the reason we don’t recommend it for warehouses with long runs.
• IT owns the network. Every switch, VLAN and firmware update follows IT’s change process.
• Egress lighting has to work during a power failure, whatever the network is doing.

What we’d do, by phase

Survey. Walk the building with the architect, the IT manager and the preservation consultant, map the chases and closets, and read the hazardous-materials report.

Bench. Build a small PoE lighting system in the Lytbit Lab: fixtures from two candidate platforms, an 802.3bt switch and the sensors. We’d measure the switch’s own consumption at idle and at full load, then reboot it mid-day to see what the fixtures do. If they go dark during a firmware update, they’re out.

Design. Fixtures and sensors run on low-voltage cable through existing chases and behind new millwork, with smaller penetrations than line-voltage branch circuits would need, subject to the AHJ’s acceptance. Switches sit in the two closets on a dedicated lighting VLAN, on a UPS sized for egress paths, with closet cooling checked for the added heat. We’d check cable bundle sizes against the electrical code’s limits for data cable that carries power. Egress lighting stays on a separately listed emergency solution unless the PoE platform’s own approach is listed and accepted by the AHJ. Sensors report presence only: no cameras, and data aggregated by room and hour.

Oversee. One schedule for the low-voltage contractor, the electrician and IT, and a written responsibility split: IT owns switches and firmware, facilities owns schedules and sequences, and we design, commission and monitor. Our cybersecurity review goes to IT before anything is connected.

Commission. Test every sequence, then test failures: pull a switch’s uplink, power-cycle a switch, and cut utility power to confirm egress behavior. The results go to IT in writing.

Operate. Twelve months of occupancy analytics in Lytbit Ledger, reported by floor and room and timed to land before the sublease decision.

Projected outcome

Measure	Current operation	Projected
Annual lighting energy, including switch overhead	about 80 MWh	32–40 MWh
Share used by PoE switches and standby	none	6–10% of the projected total
Space-use data	none	presence by room and hour
Lighting network ownership	nobody	IT, under its change process

The PoE overhead is real: a GSA trial found PoE lighting traded lower installation cost for higher standby power. We count it inside the total rather than footnoting it. Most of the savings come from replacing fluorescent fixtures and adding sensing and trim at every fixture; PoE is what makes that practical behind plaster.

What we’d watch out for

• Firmware day. Patch windows have to be coordinated with lighting schedules, and fixtures must hold their state when a switch reboots.
• Switch power budgets. A switch’s total PoE budget can be smaller than the sum of its port ratings.
• Analytics creep. Once utilization data exists, someone will ask for more of it. We’d write the privacy rules before go-live.
• Surprises behind plaster, which can move a cable route by a whole wall.

What we’d do differently next time

We’d put IT and the preservation consultant in the same first meeting, not consecutive ones. Their constraints collide in the closets and chases, and it’s cheaper to discover that at a whiteboard. We’d also publish the switch-overhead measurements from the bench as a short Bench Note before design, so the owner sees the standby cost on paper before choosing PoE rather than after.