Daniel's been chewing on something from a past conversation about MROs — those maintenance, repair, and overhaul operations that keep aircraft flying. We talked about the sheer scale of inventory at places like Delta TechOps and Lufthansa Technik, hundreds of thousands of SKUs, and he wants to crack open the actual physical side of it. Not the procurement strategy or the triage logic this time, but the literal storage. The bins, the racks, the gear that moves a two-hundred-pound actuator from a shelf in the back corner to a mechanic's bay three buildings over. And he wants to know how you handle the heavy stuff specifically, because a lot of these parts aren't exactly something you toss in a shopping cart.
The first thing to understand is that an MRO warehouse isn't one warehouse. It's about six different types of storage facility sharing a roof, and they operate on completely different physical principles. You've got your high-bay automated storage and retrieval systems for small fasteners and consumables, your vertical lift modules for medium components, your cantilever racks for long structural pieces, your pallet racking for engine stands and landing gear, and then the temperature-controlled cages for composites and sealants. And none of them use the same retrieval equipment. That's the part people miss — the inventory management software might see one unified catalog, but the physical reality underneath it is a patchwork of incompatible systems that each solved a different problem at a different time.
So the unified inventory number is a lie, is what you're saying.
It's a useful lie. The system tells a mechanic that a thrust reverser actuator pin is in stock and available, and that's true. What it doesn't tell you is that getting it from its shelf to the bay door requires a narrow-aisle reach truck that only two people on the shift are certified to operate, and one of them is on lunch.
The database says yes and the forklift says maybe.
Let me walk through the actual zones, because the storage method is dictated almost entirely by the physical geometry and weight of what's being stored. At the small end, you've got your C-class items — fasteners, o-rings, washers, cotter pins, the little stuff. These live in automated vertical carousels or vertical lift modules. A vertical lift module is basically a giant enclosed elevator with trays. You punch in a part number, the machine brings the right tray down to an access opening at waist height, a light points at the exact bin, you pick the quantity, and it closes back up. The whole unit might be forty feet tall and occupy a footprint of about a hundred square feet, but it holds something like sixty-five hundred individual line items.
And the machine is doing the walking, not the picker.
Right. The picker never moves. That's the key metric they optimize for — picker travel time. In a traditional static shelving setup, the worker spends maybe sixty percent of their shift walking. In a goods-to-person system like a vertical lift module, that drops to near zero. You can pick something like two hundred and fifty lines per hour per operator, versus maybe forty in a walk-and-pick setup. For an MRO handling a hundred thousand SKUs of small parts, that difference is existential.
So the little stuff's solved. The heavy stuff is where it gets interesting.
The heavy stuff is where you stop thinking about storage and start thinking about rigging. Let's take a main landing gear strut. This is a component that weighs somewhere between fifteen hundred and four thousand pounds depending on the aircraft type. It's not sitting on a shelf. It's mounted on a custom contoured cradle welded to a steel stand, and that stand sits on the floor of a pallet-racking bay with aisles wide enough for a fifteen-ton forklift to maneuver. The fork spacing is non-standard because the cradles are non-standard. Each assembly type has its own dedicated handling fixture, and those fixtures themselves are inventoried items. You lose the fixture, you functionally lose the part because you can't move it safely without the fixture.
Wait. The fixture has its own SKU.
It does. And it's often a custom-engineered item that was fabricated twenty years ago by a shop that no longer exists, from drawings that may or may not have been digitized. There are MROs that have a full-time engineer whose job is reverse-engineering lost or damaged handling fixtures from photographs and surviving parts. The fixture problem is one of those second-order inventory headaches that nobody outside the industry ever thinks about.
So the part's in stock, the fixture's in stock, the forklift's available, the certified operator's not on lunch. What's actually moving the thing across the facility?
This depends on distance and weight. Inside the warehouse itself, for anything under about three thousand pounds, you're typically using counterbalance forklifts — standard industrial trucks, but with some aviation-specific modifications. Soft forks, for one thing. The fork tines are often wrapped in brass or fitted with rubber boots so they don't gouge a million-dollar component. For heavier items or longer runs, you'll see heavy-duty flatbed carts pulled by electric tugs. A tug is basically a motorized walk-behind unit that can pull a train of carts weighing up to maybe twenty thousand pounds total. The operator walks alongside it with a throttle control. It's slow, it's quiet, and it can navigate through doorways that a forklift can't.
And between buildings?
Between buildings is where it gets specialized. Delta TechOps in Atlanta, for example — their main campus covers something like two point seven million square feet of hangar and shop space spread across multiple buildings. They move major assemblies on what are essentially airport tugs pulling modified flatbed trailers. These are the same tugs you see pulling baggage carts on the ramp, but they're running on internal roadways within the facility. The trailers have air-ride suspension because some components — avionics, composite structures — are shock-sensitive. You hit a pothole with a solid-suspension trailer carrying a radome, and you've just turned a half-million-dollar part into a very expensive planter.
Air suspension on a warehouse cart. That's... I mean, it makes complete sense once you say it, but I would never have thought of it.
The shock-sensitivity thing ripples through the entire handling chain in ways that seem excessive until you see the price tags. Engine stands have vibration data loggers on them now. If a stand gets jostled beyond a certain threshold during transport, it logs the event with a timestamp and GPS coordinates, and the receiving inspector knows to check for damage before the engine goes anywhere near an aircraft. Some MROs require that any ground transport of an engine over a certain value — typically above about five million dollars — gets a dedicated escort vehicle and a pre-driven route. They literally send someone out in a car to drive the path first and check for construction, potholes, low branches, anything that could cause a problem.
So you've got a route survey for moving a part across a parking lot.
For an engine, yes. And that's not even unusual. Lufthansa Technik has facilities where engine transport between buildings goes through dedicated tunnels to avoid weather exposure. The tunnels are climate-controlled. The engines travel on automated guided vehicles that follow magnetic tape embedded in the floor. No human driver, no forklift, just a thirty-million-dollar engine gliding along at walking speed on a robot cart, through a tunnel, in the dark, because nobody needs to be there.
That's a very specific image.
The automation side is accelerating, actually. The newest MRO warehouses are moving toward what they call "dark stores" for the small and medium parts — fully automated facilities where no human enters the storage area at all. Robots pick, robots pack, robots deliver to a pickup window at the edge of the zone. The only time a human touches the part is at final kitting, when everything for a specific maintenance task gets assembled into a single delivery. The business case is straightforward: automation eliminates picking errors, and in aviation, a picking error isn't a customer returning the wrong size shirt. It's a mechanic installing a fastener with the wrong heat treatment on a flight control surface.
What's the error rate on human picking in these environments?
For manual pick from static shelving, it runs about one to three percent depending on the complexity of the parts and how tired the picker is. That sounds low until you multiply it by the number of parts moved per shift. A busy MRO might process twenty thousand line items a day. At a one percent error rate, that's two hundred wrong parts pulled every single day. Most get caught at subsequent checks — the mechanic notices it doesn't fit, the inspector catches it, the system flags a weight discrepancy — but the layers of checking are there precisely because the error rate is nonzero and the consequences are catastrophic. Automation can push the error rate below zero point one percent. That's not just efficiency, it's safety engineering disguised as logistics.
So the dark store is a safety feature.
Fundamentally, yes. And the physical storage equipment reflects that philosophy all the way down. Take bin design. In a consumer warehouse, you might see cardboard bins or plastic totes, whatever's cheap. In an MRO, small-part bins are often electrostatic-discharge-safe — ESD-safe — because some avionics components can be destroyed by a static shock you can't even feel. The bins are color-coded by storage condition. Blue for standard, red for ESD-sensitive, yellow for time-limited — items with a shelf life like sealants and adhesives that cure or degrade over time. The bin itself is part of the quality control system.
The bin is a QC instrument.
It is. And the shelving those bins sit in is often louvered panel — we've talked about that before — but in MRO applications, the panels are typically grounded. There's a copper grounding strap running down the back of every upright, bonded to the building's grounding grid. That's not a recommendation, it's an FAA requirement for any facility storing ESD-sensitive aircraft parts. The shelving isn't just shelving. It's part of the electrical system.
Alright, let me pull on a thread. You mentioned time-limited items. How do they physically manage shelf-life expiration across a hundred thousand SKUs?
This is where the storage system and the inventory management system have to talk to each other constantly. Every time-limited item is received with a manufacture date and an expiration date logged into the system. The physical storage location is irrelevant to the expiration logic — the system tracks it by batch, not by bin. When a batch is approaching expiration, the system generates a pull order. A picker goes and retrieves every unit of that batch from wherever it's physically stored, and it moves to a quarantine cage. The quarantine cage is a literal locked cage, usually painted red, physically separated from active inventory. Nothing leaves quarantine without a disposition — either it gets recertified through testing, it gets returned to the manufacturer, or it gets destroyed.
And "destroyed" for a sealant tube means what, exactly?
It means documented destruction. You don't just throw it in the dumpster. There's a process, there's a form, there's a witness signature. For some items, the manufacturer requires photographic evidence of destruction before they'll issue a replacement credit. I've heard of cases where expired composite prepreg material — that's the carbon fiber fabric pre-impregnated with resin — has to be cut into pieces and photographed on a specific mat with a dated card in the frame, like a hostage photo, before the manufacturer accepts the claim.
A hostage photo for expired fabric. The aviation industry is something else.
The stakes are just different. If a tube of expired sealant gets used on a cabin door seal and it fails at altitude, that's a rapid decompression event. The cost of over-engineering the inventory control is trivial compared to the cost of getting it wrong once.
Let's go back to the heavy side for a minute. You mentioned engine stands with vibration loggers. What's the actual mechanical process of pulling, say, a CFM56 engine — that's the one on the 737, right? — from storage and getting it to a maintenance bay?
CFM56, yeah, about four thousand pounds fully dressed. The engine arrives at the MRO either still mounted on its shipping stand — that's the fixture it traveled on from the airline or the previous facility — or it gets transferred to a shop stand immediately upon receipt. The shop stand is a different beast from the shipping stand. It's designed for maintenance access, with rotation capability so the mechanics can work on the engine from multiple angles. The transfer from shipping stand to shop stand requires an overhead crane, typically a bridge crane with a capacity of ten tons or more. The crane operator and the ground crew work together using standardized hand signals — radios are sometimes used but hand signals are the primary method because they don't fail. You've got a rigger on each side of the engine, a crane operator in the cab, and a supervisor watching the whole thing. The lift itself might take twenty minutes for a job that actually moves the engine four feet horizontally.
Twenty minutes for four feet.
Because every movement is planned, called out, and verified. The lift plan is written down. The weight and center of gravity are confirmed from the engine's documentation before the sling is attached. The slings themselves are serialized and have their own inspection records — slings get retired after a certain number of lifts or if they show any sign of wear. A sling failure during an engine lift isn't a damaged part, it's a fatality. So the pace is deliberately, almost absurdly, slow.
And this is all happening inside a facility that also has robots picking o-rings in the dark.
That's the contrast that makes MRO logistics so fascinating to me. You've got Industry 4.0 automation running alongside processes that haven't fundamentally changed in fifty years, because when the object weighs two tons and costs more than the building it's stored in, you don't innovate on the lift procedure. You innovate around it — better documentation, better sensors, better planning software — but the actual physical act of rigging and lifting remains stubbornly analog.
What about the mobile equipment fleet itself? The forklifts, the tugs, the trailers. That's a whole secondary inventory to manage.
It is, and it's a significant operational burden. A large MRO might have a fleet of sixty or seventy material handling vehicles. Each one has its own maintenance schedule, its own inspection requirements, its own battery charging or refueling logistics. Electric forklifts need battery change-out stations with overhead hoists because the batteries weigh a couple thousand pounds themselves. The battery room is a hazardous area — hydrogen off-gassing during charging means you need ventilation, spill containment, and no ignition sources. It's a miniature chemical plant tucked into the corner of the warehouse.
And if a forklift goes down, the whole chain stops.
It can. There are critical-path vehicles where a single failure blocks a whole workflow. The reach truck that services the high-bay racking in the structural parts zone — if that's down for a day, you're not shipping any flap track fairings or leading-edge slats until it's fixed. So the fleet maintenance team keeps a critical spares inventory for the material handling equipment itself. Tires, hydraulic hoses, controller boards, drive motors. You've got inventory to manage the inventory management equipment.
The recursion problem. How many layers deep does that go?
It goes until someone decides the cost of the next layer of spares exceeds the cost of accepting downtime. That's a real financial calculation that the ops director makes. A replacement drive motor for a specialized narrow-aisle lift truck might be twelve thousand dollars and take eight weeks to arrive. Do you stock one? It depends on the probability of failure, the cost of the downtime it would cause, and whether you can work around the failure by rerouting picks to a different zone. Some MROs stock them, some don't, and both decisions are correct for their specific circumstances.
Daniel asked about Delta and Lufthansa Technik specifically. Do they approach this differently?
They do, and it reflects their different business models. Delta TechOps is an in-house MRO that also does third-party work — about eighty percent of their business is actually for other airlines now. Their parts operation in Atlanta is built around speed. They hold an enormous inventory on-site because they're supporting Delta's own fleet, which means an AOG — aircraft on ground — situation for a Delta aircraft gets priority access to anything in the building. Their storage philosophy is "have it, have it close, have it accessible." Lots of forward-stocking locations near the hangar bays, almost like a hospital having supply closets on every floor instead of one central pharmacy.
Forward-stocking meaning they duplicate inventory across multiple locations.
Yes, deliberately. The same part number might exist in the main warehouse, in a satellite stockroom near the narrow-body hangar, and in a mobile kit that travels with a specific maintenance team. The system tracks all of it as one logical inventory pool, but physically it's distributed. The tradeoff is higher total inventory carrying cost in exchange for lower mean time to repair. For an airline, minutes of downtime are so expensive that the carrying cost is trivial by comparison.
And Lufthansa Technik?
Lufthansa Technik is more of a third-party shop with a global footprint. They've got facilities in Europe, Asia, the Americas, and they're servicing dozens of different airline customers with different fleet types. Their inventory challenge is breadth, not just depth. They can't forward-stock everything for everyone, so they rely more heavily on their logistics network to move parts between facilities. Their storage design emphasizes consolidation and cross-docking — parts arrive, get processed, and get shipped out to wherever they're needed, often without spending much time in static storage. The physical layout of their warehouses reflects that flow-through philosophy. Receiving and shipping are adjacent, with minimal deep-storage between them.
So Delta's model is a library with books on every floor. Lufthansa's is a distribution hub where most books are just passing through.
Both models require the same fundamental physical infrastructure — the racks, the bins, the forklifts, the cranes — but arranged in completely different layouts because the operational philosophy is different. The equipment catalog is similar; the floor plan is not.
What's the single heaviest thing that moves through one of these facilities routinely?
A fully assembled engine, no question. But the second-heaviest routine move is probably a landing gear set. A 777 main landing gear assembly — that's the whole truck, the strut, the actuators, the wheels and brakes still mounted — can weigh over eighteen thousand pounds. Moving that requires a dedicated transport trailer, often a multi-axle lowboy style trailer pulled by a heavy tractor unit. The floor in the landing gear shop is reinforced concrete, often eighteen inches thick or more, with embedded anchor points for securing the assembly during teardown. The building is literally designed around the weight of the parts it houses.
Eighteen inches of concrete so the floor doesn't crack under a landing gear.
And the anchor points are load-rated and tested periodically. If you walk through a landing gear shop, you'll see load-rating plaques on the floor, same as you'd see on a bridge or a crane. The floor is an engineered structure.
I'm trying to picture the transition from the precision automated small-parts area to the landing-gear bay. Are they even in the same building?
Sometimes yes, sometimes no. At the largest facilities, the heavy bay is essentially an aircraft hangar that happens to be used for component storage and teardown rather than whole-aircraft maintenance. It has the same door height, the same overhead crane capacity, the same floor loading as a maintenance hangar. The small-parts automation lives in a completely different building envelope — lower ceiling, tighter climate control, cleaner environment. The two zones are connected by the internal roadway network and the tug-and-trailer system. The parts don't mix physically until they reach a kitting area, which is typically at the edge of the heavy bay, where the small consumables and the large structural components get consolidated for a specific job.
The kitting area is the handshake point between the two worlds.
And kitting itself is a whole discipline. For a major maintenance event — say a C-check on a wide-body aircraft — the kit might include three thousand individual line items. Everything from cotter pins to a flap actuator. The kit is assembled over several days before the maintenance event starts, staged in a dedicated holding area, and then delivered to the hangar bay as a single unit. The kit cart itself is a custom-designed rolling platform with multiple levels, labeled sections, and a master inventory sheet. The mechanic working on the aircraft doesn't go to the stockroom. The stockroom comes to the mechanic, pre-organized, pre-verified, and sequenced in the order the job card calls for.
So the mechanic never sees the warehouse at all.
Ideally, never. The warehouse exists to make the warehouse invisible to the person turning the wrench. That's the whole design philosophy. Every piece of storage equipment, every automated system, every tug and trailer and crane — it's all in service of making sure the right part appears at the right bay at the right time, and the mechanic's mental model of the operation doesn't need to include any of it.
That's a... there's something almost beautiful about that. The entire logistical apparatus is a stage crew, and the mechanic is the performer who never sees the ropes and pulleys.
And when it works, it's invisible. You only notice the logistics when they fail. When a mechanic waits forty minutes for a part that should have been kitted, suddenly the warehouse is very visible and everyone is very unhappy.
Hilbert: They're right.
Hilbert.
Hilbert: The kitting is the whole thing. I worked a summer in a warehouse that supplied a truck assembly plant in Ohio, nineteen ninety-three. My job was picking fasteners into plastic bags for the line-side kits. Fourteen thousand steps a day on the pedometer, which I only know because my supervisor wore one and made me wear one too. He was convinced step count correlated with productivity. It doesn't, by the way. The fastest picker on the shift was a guy who figured out he could combine three trips into one by loading his cart in a specific sequence and he cut his steps by forty percent and his pick rate went up. Management tried to make everyone copy his method and most people couldn't do it because it required remembering about sixty part numbers in your head.
The mental map problem. That's exactly what the vertical lift modules solve.
Hilbert: They do now. Back then it was a guy named Rick with a freak memory and management thought it was a process they could document. They wrote a twenty-page SOP and nobody read it. The real lesson was that good picking is a cognitive skill, not a physical one, and the industry spent the next thirty years building machines to remove the cognitive load. Which worked. But the thing you two didn't mention — and it's the part that actually keeps me up — is what happens to the kit after it's delivered.
After it's delivered to the bay.
Hilbert: The mechanic opens the kit, uses about sixty percent of what's in it, and the other forty percent goes back. The return loop. The reverse logistics of unconsumed parts. And the error rate on returns is about four times the error rate on picks, because the mechanic is thinking about the torque sequence on a fuel pump, not about whether the three unused washers go back in bin four-C or bin four-D. So you get cross-contamination of inventory. A washer that looks identical but has a different heat treatment ends up in the wrong bin. Six months later someone picks it for a different job, installs it, and now you've got a latent defect that won't show up for thousands of flight hours.
The return loop. That's... I focused entirely on the outbound side. You're right, the return is where the real contamination risk lives.
Hilbert: Every MRO I've ever talked to has a return-inspection process, and every one of them will tell you it's the weakest link in their quality system. The parts are small, they look the same, the markings wear off, the paperwork gets separated. Some places have gone to a no-return policy for certain consumables — if it leaves the stockroom, it doesn't come back, period. The cost of scrapping unused fasteners is less than the cost of one misidentified washer in the wrong engine. But that's a hard sell to the finance department because it looks like waste on a spreadsheet.
It's not waste, it's insurance.
Hilbert: Try explaining that to a cost accountant who's never set foot in a hangar. I've had that conversation. It doesn't go well.
How do the automated systems handle returns? In a dark store setup, there's no human to make the judgment call.
Hilbert: They don't handle returns at all, mostly. The automated zone is outbound-only. Returns go through a separate manual inspection and restocking process, and only after a quality inspector has verified every piece. The automation is fast because it only has to solve half the problem. The other half is still a person with a magnifying glass and a material certification database.
And that person is probably understaffed.
Hilbert: Always. It's the least glamorous job in the building and it's the one that catches the mistakes that would otherwise end up in an NTSB report. I used to know a woman who did that job at a facility in Wichita for twenty-two years. She could tell the difference between a cadmium-plated washer and a zinc-plated washer by the smell. Not kidding. She retired and they replaced her with two people and a spectrometer and the error rate still went up.
The spectrometer can't smell.
Hilbert: It cannot. And the woman's name was Dolores and she trained both of her replacements and then they still called her at home for the first year. She didn't mind. She said it was nice to be needed.
There's a whole episode in Dolores. But you've got me thinking about that return loop now. The kitting process Herman described — the kit arrives at the bay, the mechanic uses what they need, the rest comes back. If the return is the weakest link, then the whole just-in-time kitting philosophy has a built-in failure mode that nobody's fully solved.
They've solved it partially with single-use packaging. Some consumables now come in sealed kits where the kit itself is the unit of issue — you open it, you use everything inside, you dispose of the packaging. No partial returns because there's nothing to return. But that only works for standardized job packages. For the unpredictable stuff — the "oh, we found corrosion here, now we need these additional parts" — you can't pre-package everything.
So the system is optimized for the predictable and vulnerable at the edges where the unpredictable happens.
Which is also where the most critical repairs happen, because the predictable stuff is routine and the unpredictable is where you find the problems that weren't supposed to be there.
The misconception people have about MRO warehousing, I think, is that it's just a bigger version of a Home Depot back room. More shelves, more bins, more forklifts. And what we've actually described is a facility where the shelving is grounded to the electrical grid, the bins are color-coded QC instruments, the floor is an engineered load-bearing structure, the trailers have air suspension, and the whole thing exists to be invisible to the person it serves.
The misconception is that scale is just more. Scale changes the category. A warehouse with a thousand SKUs is a storage problem. A warehouse with a hundred thousand SKUs where picking errors can kill people is an entirely different discipline. The physical equipment reflects that — it's not consumer shelving writ large, it's industrial process equipment that happens to hold inventory.
The one thing I keep coming back to is the fixture problem. That somewhere in a building in Atlanta or Hamburg there's a custom steel cradle for a part they haven't ordered in eight years, and if that cradle gets lost or damaged, the part becomes effectively unavailable even though it's sitting right there on a shelf. The physical storage is only as good as the handling equipment that surrounds it.
And that handling equipment is often irreplaceable. You can't order a new fixture from a catalog. You have to re-engineer it from the part it was designed to hold. It's the inventory problem one level removed — the inventory of things that hold the inventory.
Daniel's going to want to know about the euro boxes. I know he is.
They're in there. The small-parts bins in the vertical lift modules are often standardized containers — not always euro boxes specifically, but the same principle. Standardized modular containers that fit standardized shelving that fits standardized automated systems. The standardization is what makes the automation possible. If every bin were a different size, the robot couldn't grab them.
So his garage system scales to aviation, is what you're saying.
The principles scale. The specific container dimensions are different — aviation uses a lot of proprietary bin sizes designed around specific shelving systems — but the modularity concept is identical. Daniel's euro box obsession is, I hate to admit it, well-founded.
I'm not telling him you said that.
He'll hear it anyway. He listens to every episode.
He does. And he'll probably email us about the return loop within the hour.
Good. It's worth a whole episode on its own.
Something to come back to. For now — this has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop, who apparently spent a summer in nineteen ninety-three learning lessons about warehouse picking that the aviation industry is still wrestling with.
Find every episode at my weird prompts dot com. If you've got a question about the physical guts of an industry most people never see, email the show at show at my weird prompts dot com. We'll be back soon.
