From 8 to 24 Rockets Per Year: How a Tier-1 Aerospace Supplier Unlocked Launch Vehicle Production Capacity

The production manager stood in the machine shop at 6 AM on a Tuesday in 2024 and realized the operation was running blind. Seventeen CNC machines, two massive welding stations, and a composite autoclave were generating data streams that nobody was actually reading. When a tolerance issue surfaced on a turbopump casing, the team found out three days later, when the part arrived at assembly. By then, twenty-two identical castings had already moved through the same machine with the same offset.

This was a Tier-1 aerospace supplier embedded in the US space launch vehicle supply chain. They had been holding steady at eight completed engine sections per year. Their customer needed twenty-four. Not in five years. Not in a long-term roadmap. Now. The company faced a choice: build a new facility or extract capacity from existing iron.

Challenge

Space launch vehicle manufacturing operates in a peculiar bandwidth. The production volumes are low compared to automotive. The tolerances are brutal. Rework on a single engine section can kill six weeks of schedule. The supplier's bottleneck analysis revealed something unexpected: it was not machine utilization. Machines were running 68 percent of available time, which is healthy for job shop work. The real constraint was invisible.

Root cause analysis showed that work orders were moving through the floor with no real-time visibility into quality, dimensional drift, or tool condition. A CNC machine grinding a turbine blade housing would drift 0.003 inches over a run of six parts. The part would pass inspection at the machine. The next operation, thirty-six hours later in a different cell, would find the problem during CMM verification. By then, the rework decision tree had exploded. Can you salvage it? How much stock removal? What is the schedule impact? Engineering would spend two days generating a deviation request. Meanwhile, the machine that caused the problem was working on something else, and nobody was tracking why it drifted in the first place.

The operation was also saddled with changeover dead time. Moving from a nine-axis impeller machining program to a five-axis vane cascade setup took four hours. Much of that was search time: finding the right fixturing, hunting through digital archives for the program variant, validating tool offsets on machines without automated probe systems. With only eight engine sections per year, changeovers happened infrequently enough that tribal knowledge evaporated between runs.

Solution

The supplier did not install a new ERP or buy a lights-out automation system. They installed what amounts to a nervous system: spindle-level data collection and a manufacturing execution dashboard that surfaced anomalies in real time. Every CNC machine got a controller-integrated telemetry module. Accelerometers on spindles flagged tool wear before catastrophic failure. Temperature and humidity logging in the composite bay caught oven ramp issues before they baked resin delamination into structure. CMM machines began streaming dimensional data back to a central repository instead of sitting in isolation with USB drives and notebooks.

The second move was process standardization, which sounds dull but was radical here. They documented changeover sequences in video, step by step, with actual time stamps. They built fixturing kits in aluminum cases, labeled and inventoried, ready to grab. They indexed the CNC program library and added metadata tags: part family, tool path type, material, machine compatibility. What had been a custom engineer hunt became a thirty-minute retrieval and setup procedure.

The third piece was a weekly rapid-cycle manufacturing review. Every Monday, the team looked at the previous week's spindle data, tool breakage events, and dimensional trends. If a machine had drifted on Friday, they understood why by Monday morning and had implemented a fix before the next run. This is not novel; this is standard work. It required discipline and time, but not technology beyond what they already owned.

Results

In eighteen months, the operation moved from eight engine sections per year to twenty-four. Schedule reliability moved from 73 percent on-time delivery to 94 percent. Rework, which had consumed roughly nine percent of labor hours, dropped to 2.1 percent. Changeover time fell from four hours to sixty-eight minutes. They added one additional shift and converted one general-purpose machine to dedicated impeller work, but no facility expansion.

The customer increased orders again in Q4 2025. The supplier is now at thirty-two engine sections annually and is already optimizing the composite cure cycles to unlock the next constraint. The lesson, unstated but visible on the shop floor, is this: capacity lives in visibility. The machines were always there. The engineers were there. What was missing was knowing in real time what the machines were telling them.