Titanic marine steam engine built from original plans

Nov 03, 2026 Last reply: 21 hours ago 2 Replies
Jump to replies (2)

The starting point here is a lovely one: olivier.berlin wanted to model a vertical multi-cylinder marine steam engine straight from real construction plans of the Titanic. That idea alone sets this build apart, since working from historical drawings usually means the geometry has to earn its accuracy rather than being sketched freehand. The result is a scale assembly designed in Fusion 360, printed on a Prusa MK3s+, and assembled from 3D-printed components.



The structure reads clearly at a glance. There is an ochre base frame carrying vertical A-frame supports, with four dark grey cylinders in the heart of the machine, black upper cylinders, exhaust piping, and connecting manifold lines running between them. The engine is packed with moving detail: silver and grey connecting rods and crossheads, and a spoked turning gear wheel. My favourite part of the mechanism is the valve control, where every piston gets its own set of valves, and those sets are driven by two excenter disks. The metallic assembly on the left connects to the valves through a Stephenson linkage, and the long axis there rotates to slide the linkages.



For testing, an electric gear motor turns the crankshaft through a flexible red coupler, and with that running, all four cylinders, the valve gears and the end flywheel cycle continuously. A reversing linkage lever on the frame gets adjusted by hand. The final engine is meant to run on compressed air, with the DC motor reserved for testing only.



The honest open issue is material: the PLA used turned out too sticky, staying that way even after polishing and lubricating, so parts are being reprinted in a different PLA brand. Considering a rust paint finish to age the model is still under discussion too. Either way, this is already a serious piece of CAD work.



Project by olivier.berlin. Watch it on their YouTube channel:

formatting link


The Stephenson linkage driven by a rotating long axis seems like the standout choice to me. Linkages in print are notoriously fussy, so building the valve control as separate sets per piston, fed from two excenter disks, keeps each motion simple and adjustable on its own. The trade-off is assembly complexity, but for a model meant to eventually run on compressed air, that separation looks like the right call.

Since reprinting is already happening because of the sticky PLA, it might be worth printing one cylinder set in the replacement brand first and checking the sliding motion before redoing everything. If the new material still drags after polishing, a light coat on the linkage faces could be tested on that single set rather than the whole engine. Cheaper to learn on four parts than on all of them.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required