The True Installed Cost of Fixture Replacement Versus Retrofit in a Live Process Unit

Lumens Per Watt Is the Wrong Opening Question

Energy is a real saving and it is almost never the deciding one. Run the arithmetic honestly on a single 400W metal halide fixture:

• Input power with ballast losses: roughly 455W, not 400W
• 4,000 operating hours per year, $0.09/kWh → about $164/year
• Replace with a 150W LED head → about $54/year
• Net energy saving: ~$110 per fixture per year

Over a ten-year horizon that’s $1,100 per fixture. Meaningful, not decisive — and it sits on the same order of magnitude as one access event. Which means the entire economic argument for hazardous area LED lighting is decided by installation and maintenance logistics, not by watts. Anyone building a business case on energy alone is optimizing the small variable.

Anatomy of a Full Replacement in a Live Unit

Here is a realistic cost stack for replacing one fixture at ~25 feet on structure inside an operating Class I, Division 2 area. Treat these as illustrative ranges — every site’s loaded craft rates, contractor structure, and permit culture differ — but the shape of the stack is consistent across refining, midstream, and chemicals.

The fixture is 15–20% of the job. Everything else is the event.

Permits Are a Throughput Constraint, Not a Line Item

A fixture swap in a classified area can trigger four permit systems at once: energy isolation under 1910.147, an energized electrical work permit or established electrically safe condition per NFPA 70E, work at height, and — the expensive one — a hot work permit under NFPA 51B and OSHA 1910.252.

Hot work is the hinge. Grinding a seized fastener, cutting a corroded nipple, applying torch heat to break a galled joint — all of it is hot work, all of it requires gas testing and a fire watch, and in many operating units hot work in a classified area is either heavily restricted or flatly prohibited while the unit is running. Whether you anticipate hot work determines whether this job is a Tuesday or a turnaround.

There’s a second-order effect estimators consistently miss: permits are rate-limited. A unit may issue a fixed number of hot work permits per day. Lighting competes against mechanical, inspection, and instrumentation work for those slots, and lighting loses. Your crew’s productivity is capped by the permit desk, not by their skill.

Access Is the Biggest Line, and It’s Identical Either Way

This is the single most important number in the estimate, and it deserves to be said plainly: access costs the same whether you replace the fixture or retrofit it.

You still need the scaffold. Engineered scaffold in a live unit means design, erection permit, tagging, competent-person inspection per OSHA 1926.451, and dismantle. A single tower to reach one fixture is frequently $600–$1,500 all-in, and that assumes the deck below is clear — which, on a congested pipe rack, it often isn’t.

The alternatives each carry their own friction. A mobile elevating work platform needs ground access that process congestion may not permit, and a diesel unit inside a classified area needs spark arrestors, continuous gas monitoring, and its own permit. Rope access via SPRAT- or IRATA-certified technicians is often the fastest route to steel that scaffolding can’t economically reach, but requires anchor engineering and a two-person minimum with rescue capability.

Because access is common to both options, it is not where retrofit saves money. Retrofit saves below the access line — and it saves schedule, which is worth more.

Conduit Seals: The Skilled Work Nobody Budgets

Under NEC Article 501.15, sealing fittings in Class I locations must be installed within 18 inches of the enclosure, and boundary seals are required where conduit leaves the classified area. Pull a fixture and you very often break a seal.

Rework is not trivial. You chip out cured compound, verify or replace conductors through the fitting, install a fresh fiber dam, mix and pour new sealing compound, and then wait. Cure time is dead time on a permit clock, and the fitting itself may not be reusable after 25 years of thermal cycling and corrosion. Call it $200–$450 per seal in materials and skilled labor, plus the schedule drag — and every reopened seal is a new inspection point in an area where the seal is a fire-safety barrier, not paperwork.

A retrofit performed through the fixture’s own access cover disturbs no seal at all.

The 25-Year Threaded Joint

Now the honest one. Explosionproof construction requires five full threads of engagement, and that joint has been in place since the Clinton administration.

What happens when you put a pipe wrench on it:

• It comes loose cleanly. Roughly the outcome you planned for.
• It’s galled or galvanically seized. Aluminum housing against galvanized rigid conduit in a coastal or acidic-service environment fuses over decades. Now you need heat, penetrant, and time — or you cut, and cutting is hot work.
• The nipple snaps. You’re now replacing a conduit body, possibly a run.
• You find water, corrosion products, or degraded conductors inside. One fixture just became a circuit replacement, and if the wiring predates the plant’s current standards you may inherit an insulation upgrade you never scoped.

The point isn’t that this happens every time. It’s that it happens on some meaningful fraction of joints — and a single fixture that turns into a conduit run is a $1,500–$4,000 surprise in the middle of a permitted window with a crew already mobilized.

Disturbing a 25-year-old threaded joint is the highest-variance decision in the job. A retrofit doesn’t make that decision at all.

The Constraint That Actually Decides This: Turnaround Access

Here’s where the conversation stops being about cost and starts being about calendar.

If your fixture replacement program requires hot work in classified areas, it is a turnaround scope item. That has consequences that dwarf the unit cost:

• Turnaround scope is ranked, and lighting sits near the bottom. Heat exchanger bundles, tower internals, catalyst change, and inspection findings all outrank it. Lighting is the first scope cut when TAR hours compress.
• The interval is the interval. If the unit runs on a four- or five-year cycle and lighting gets bumped once, you’ve just added five years to your program.
• Turnaround day rates and craft premiums are materially higher than routine maintenance rates.
• Deferred lighting compounds: as HID fixtures fail between turnarounds, illuminance degrades, and you accumulate dark areas that create their own safety and operability exposure — and reactive one-off replacements at full access cost, which is the worst unit economics available. Lamp availability is degrading on its own schedule too, which compresses the window further.

A retrofit that requires no hot work, no seal break, and no conduit disturbance can frequently be executed on-stream, by in-house crews, on a normal work order. That is not a 30% cost saving. That is the difference between doing the work in Q2 and doing it in 2030.

For a maintenance manager, that’s the whole argument.

The Number Nobody Runs: The Avoided Relamp

One more piece the energy model omits.

An HID fixture doesn’t sit untouched for ten years. Group relamping on a three-to-four-year cycle is standard practice, and every relamp requires the same permits, the same isolation, and — critically — the same scaffold. Three relamp events over a ten-year horizon at $700–$1,200 of access-and-labor apiece is $2,100–$3,600 per fixture in maintenance events that a long-life LED simply eliminates. That also assumes you can still buy the lamp in year seven, which is an increasingly unsafe assumption.

Set that against the $1,100 of ten-year energy savings and the ranking becomes obvious:

Avoided maintenance access is worth roughly two to three times the avoided energy.

If you build one slide for the capital request, build that one.

Batching, Mobilization, and Why Unit Price Misleads

Access cost amortizes; fixture cost doesn’t. That single fact should drive execution strategy.

One scaffold set that services eight fixtures on the same rack section spreads $1,200 of access across eight units — $150 each instead of $1,200 each. Mobilization, permits, gas testing, and pre-job briefings amortize the same way. This is why a planned campaign against a mapped fixture population beats reactive replacement by a margin that has nothing to do with which fixture you bought.

Practical implication for estimating: stop quoting per-fixture installed cost as a constant. Quote it as a function of cluster size. A well-batched program might land at $900–$1,600 per fixture installed; the same fixtures replaced one at a time as they fail will run $3,000–$6,000 each, forever.

The Honest Case for Full Replacement

Retrofit isn’t the answer everywhere, and pretending otherwise costs credibility with exactly the audience you’re trying to reach.

Replace the whole fixture when:

• The housing is at end of life. Corroded body, crazed or pitted globe, gasket taking a permanent set, damaged flame path surfaces. Retrofitting a new light engine into a failing pressure envelope buys you nothing.
• No certified retrofit exists for your host enclosure. If the kit’s certificate doesn’t name your specific housing by manufacturer and model — the second half of the test the 2026 NEC now spells out — you’re creating an uncertified assembly and inheriting manufacturer-level responsibility for it — the temperature classification is the clearest illustration of why that matters.
• The conduit run is already suspect. If you’re going to have to open it eventually, opening it during a planned turnaround with a full crew beats opening it during an unplanned failure.
• You need clean documentation. A new listed luminaire arrives with an unambiguous nameplate, a single certificate covering the whole assembly, and a warranty. That’s worth real money in an audit.
• The area classification or illuminance requirement has changed. New layout, new sources, new lux targets — a retrofit into the old photometric footprint may not meet the current design basis.

Bottom Line

The fixture is 15–20% of the installed cost. Access is the largest single line and it’s a wash between the two options. The real differentiators are permit class, seal disturbance, threaded-joint risk, and whether the work can happen on-stream — and those four determine both the cost and the calendar.

When you build the estimate, build it as an event, not a purchase:

1. Cluster the fixture population by access point, not by area.
2. Price access once per cluster and amortize.
3. Assign a probability-weighted scope creep contingency to every joint you plan to break.
4. Test whether the job can avoid hot work entirely. If it can, it’s on-stream work. If it can’t, it’s turnaround work — and price the five-year deferral risk explicitly.
5. Model avoided relamp events alongside avoided kilowatt-hours. The maintenance line is the bigger one.

A $640 fixture is a purchase order. Getting it installed is a project. The organizations that get this right are the ones that stopped estimating the first thing and started estimating the second — and that checked the supplier’s certification claims before either number went into a capital request.

Related reading: OSHA 1910.307 hazardous locations · OSHA 1910.147 lockout/tagout · OSHA 1910.119 process safety management · OSHA 1910.146 permit-required confined spaces · EPA universal waste lamps · 40 CFR Part 273 · NEC free access · API recommended practices