In a Dallas-Fort Worth July, the limiting factor on a mechanical repair is usually the part, not the technician. A cooling tower fan motor failed at a multi-tenant office property during triple-digit heat. A crew was available. The part was the problem: the manufacturer sold only through distributors, two suppliers in the region carried candidate components, and the matching shaft could not be confirmed or priced until the order was already placed. Expedited freight was purchased and the expedite did not happen. The shaft moved separately under its own tracking number. End to end, the building ran hot for about eight extra days, and almost none of that time was labor.
Peak season changes which constraint governs the schedule. Most of the year, the question an operator asks is how soon a qualified crew can be on site. From late June through August in North Texas, crews are still findable. Inventory is not. Everyone within a hundred miles is running long cooling cycles at the same time, regional stock draws down, and the components that keep central plant equipment turning start moving through channels that were never built for speed.
The channel, not the shelf
This motor was not a shelf item. The manufacturer sold exclusively through appointed distributors, which means the ordering path had exactly as many steps as the channel required and no shortcuts existed for a hot building. Two suppliers in the region stocked candidate parts. Two is not a market. Two is a coin flip with a lead time attached. When one of them cannot confirm the match and the other is out, the sourcing conversation stops being about price and becomes a question of who can produce the component at all.
This is worth internalizing before it happens. A component that is single-source or distributor-only behaves nothing like a filter or a belt. The ordering decision on it is slow by design, and peak season is the worst possible moment to be learning that for the first time.
Expedited freight buys a service level, not an outcome
Overnight freight was paid for. The expedite did not happen. That is not a scandal and it is not anyone's failure; it is what a freight commitment actually is. An expedite charge buys an attempt at a delivery window inside a network running peak volume. It does not move the part any faster than the network can move it, and it does not reach backward into the pick and pack process at the origin.
Which is where the second surprise came from. The order shipped as more than one piece. The shaft traveled separately, under a different tracking number, and nobody caught it until the box was opened on site with a crew standing there. The freight promise had been trusted rather than verified: one tracking number was watched, one delivery was confirmed, and the assumption that a single order equals a single shipment held right up until it did not.
The go-forward rule is simple and unglamorous. Confirm piece count, not just shipment status. Ask which tracking numbers exist and read the scan history on each of them. Re-verify the morning of the scheduled work rather than the afternoon before it. Freight is a probability, and the only defense is looking at it directly.
A part that cannot be priced until it is ordered
The shaft could not be confirmed or priced until the order had already been placed. That sentence breaks a normal approval workflow, because every approval workflow assumes a number precedes a decision. Here the sequence inverts: commit first, learn the cost second.
Pretending otherwise costs days. Waiting for a firm quote that will not exist means the order does not get placed, and the building stays hot while a purchase order waits on a figure nobody can produce. The workable answer is to authorize a not-to-exceed range for the unpriced component up front, in writing, with the reason a firm quote is unavailable stated plainly and the release authority named. Ownership is still in control of the exposure. It is just controlling a range instead of a number.
The after-hours premium was the cheapest line on the invoice
When the missing piece finally landed, a local supply house opened after hours, for a fee, so the crew could finish that night. The install ran from 5 pm to roughly 1 am. On the invoice, that fee looks like a premium for convenience. Measured against what it displaced, it was the least expensive line item on the document.
The arithmetic is worth doing out loud. The alternative was waiting for normal business hours, which meant the crew demobilized, the schedule slipped to the next available window, and the building carried another day or more of heat across every tenant in it. A one-time access fee bought back that day. Occupant impact is the expensive quantity in a cooling failure, and it is the quantity that rarely appears on any invoice, which is exactly why it gets undercounted when someone is scanning line items for the number that looks unusual.
The durable fix happens before the failure
None of the individual decisions in this sequence were wrong. The problem is that all of them were made under time pressure, at temperature, with tenants calling. The durable fix is to move the sourcing decision out of the outage and into a quiet week.
That means identifying long-lead and single-source components on the equipment that carries real occupant impact, before those components fail. Walk the central plant and the assets whose loss empties a floor. Record what is actually installed. Then ask the distributor channel one question per significant component: is this stocked in the region, and if not, what does the ordering path look like and how many pieces does a complete order contain. The output is short, a handful of assets and a handful of parts, and it converts the worst hour of a July outage into a decision that was already made in February.
The same logic underwrites the case for planned work generally. Scheduled attention surfaces the components worth thinking about ahead of time, and the cost gap between planned and unplanned mechanical work is well documented in the comparison between emergency and preventive HVAC spend. Parts logistics is the part of that gap nobody prices until they are standing in it.