Space Manufacturing Industrialization Starts in the Lab

We used to produce 1 satellite in 2 to 3 years. Whenever a problem was detected on the shop floor, the engineers would come and fix it. And it worked! Some of those satellites are still operating well past their intended life..

But in NewSpace, a single company can now ship 10s to 100s of satellites a year. If an engineer needs to come and work on every satellite, then the model is not scalable and every unit costs more.

I gave a talk this month on this topic at the SmallSat conference in Utah. In it, I discussed how teams can transition faster from building prototypes to series production. This article covers the main points

• Adopt DFMAIT (Design for Manufacturing, Assembly, Integration, and Testing) so your design carries production rate as a constraint, alongside performance and tolerance.
• Mature your AIT process in step with the hardware, so production inherits a working process at handover rather than rebuilding one.
• Digitize before you automate, so your automation budget goes to the steps that cost you the most units and hours.

Your production rate is a byproduct of design decisions

The V-cycle rewards engineers for pushing the design, but not production rate. If performance is the only metric considered during design, then production suffers. A material with higher performance but lower manufacturability (e.g., harder to integrate or test, longer to procure) might be preferred. Those choices compound. A handful of them can add tens of hours to every unit you build, and push your scrap rate up with them.

Design for manufacturing, assembly, integration and test (DFMAIT) is the discipline that closes this. In addition to performance, you also consider simplicity, reliability, and repeatability so the line can produce units faster and cheaper. You might sacrifice a little performance, but you make it back by having robust and reliable products that can be produced at a high rate.

Adopting DFMAIT means embracing agile design, the polar opposite of V-cycle:

• Quickly build a functioning prototype, characterize it, compensate, and repeat. Do not wait for perfect tolerance!
• Use parametric analysis to identify the most sensitive parameters: build prototypes with different manufacturing processes, test, compare, and iterate.
• Collect feedback from the shop floor at every iteration to improve the design.

To learn more about this agile approach, read the summary of our roundtable discussions with ThrustMe, ESA, Airbus, and EDGX: What Breaks When Space Manufacturing Scales.

Your industrialization pace is determined by engineering processes

Space hardware maturity moves in steps: Breadboard → Engineering Model (EM) → Qualification Model (QM) → Flight Model (FM) → Small Series → Large Volumes. AIT processes have to move with it.

If not, you end up in a situation where

• Customer orders are coming in
• AIT procedures are partially documented, the rest sitting in engineers head
• Lessons learned were captured on disparate files or even paper

So pressure mounts and everyone is scrambling to write down procedures and hire skilled operators to run them.

To prevent this, make these changes to your engineering processes:

• Data is one of your most critical assets for the future, so convince your engineers to centralize and structure their data in a consistent format, from the first day of design.
• Do not wait until your flight model to start documenting your processes. Encourage your engineers to build modular blocks of AIT procedures that will gradually build the future complete production flow.
• Lighten their load. Engineers are under pressure and pulled in many directions. Give them help to automate their own scripts and invest early in automation in plotting, evaluation, and reporting. This last move can save hours every week.

Digitize early to capture learnings

Some teams are tempted to jump straight to full automation. But without step-level data, you are automating on instinct, and instinct picks the steps that feel slow rather than the ones that cost the most.

Digitize first. Every AIT step gets executed and recorded: who ran it, how long it took, what was measured, what failed. You capture all the information needed by the design team to improve MAIT (Manufacturing, Assembly, Integration, and Testing) and decide where automation pays:

• Which steps generate the most non-conformities
• Which steps take the longest
• Which steps only your most experienced engineers can run

Then automate the top item, measure again, and move to the next one. Each step funds the following one, so industrialization becomes an operating expense you spread over time instead of one large capex to justify upfront.

Digitizing early also protects what your team learns. Engineers move between programs and companies. If characterization data and lessons learned sit in a shared system, they will be indexed, so easy to search and reuse for data analysis. A failure on unit 300 will be linked to data measured during the design, accelerating root cause analysis.. If they sit in personal folders, the same investigation could take days.

A few rules that make this work:

• Capture step-level timings from unit 1, even when the step is fully manual.
• Automate one step at a time, and re-measure before starting the next.
• Make the digitized procedure the only version in use. If an operator can still run a printed copy, your data has holes exactly where you need it.

The final boss is mindset

Someone in the room asked me the question I get most often: going from single satellites to thousands, where is the real bottleneck for the industry?

The answer is mindset. People have built hardware one way for an entire career, and asking them to build it differently is cultural work that no amount of software shortens. It does not appear on a Gantt chart, and it is the longest lead item in most industrialization programs.

If you want to learn more about this journey, Read the Anywaves case study. They started it with no factory and one very capable RF technician. They now design the process in one place and deploy it to 3 production sites on two continents.