Meco 0.6: Engine Assemblies (build it, see it, test fire it)
"The nozzle carves itself out of the math now. Everything upstream of it is still your problem." -- Engineer Aoyama
Meco 0.6 is live. You set the numbers that matter. The sim builds the hardware: a real Rao bell, the cooling channels, the chamber. And for everyone who asked whether Meco is all just math under the hood: this is the release where you can see it.
Engine assemblies
Drop a Thrust Chamber into your schematic, set the throat radius, expansion ratio, contraction ratio, chamber length, and section count, and the hardware materializes. The nozzle is a real Rao thrust-optimized bell, the same contour family real engines fly, not a stock curve. You set the throat radius and the expansion ratio. It works out the other four hundred cooling channels.
The Turbopump gets the same treatment: four shaft layouts, from a single shaft to a geared three-shaft, with the shafts and gears sized from the layout you pick and your choice of pump inlet orientation. Change a headline number and the hardware rebuilds.
See your engine in 3D
All of it renders live. The thrust chamber draws itself as you design it, the Rao bell, the chamber solids, the cooling jacket, in a 3D view you can pin and resize. Wander into geometry that wouldn't work, and the editor tells you where you left the Rao chart before you burn an afternoon on it.
The nozzle flow field is a thing you can look at now. The method-of-characteristics solution renders as an interactive 3D volume inside your bell, so you can spin the camera through the supersonic flow. Thrust comes from integrating that field at the mass flow your simulation actually delivered. And if you push a bell too far at sea level, the view shows it: the flow lets go of the wall, the field grays past the separation point, and thrust is counted only where the flow is still pushing.
The turbopump gets its own machinery viewport: shafts, gears, pumps, and turbine stages drawn from the layout you picked, with casing ghosting and flange views so you can see how it all packs together.
Design the cooling jacket
Regenerative cooling gets its own drafting table. Choose how many sections run coolant and the channels tile themselves around the wall. The rib land you set drives the channel width, section by section, in a table you can read like a drawing, and the channels taper along the chamber where you tell them to. There's a cross-section view, so you can see the wall you're building.
Test-fire your own Merlin 1D
The Merlin 1D, the engine that flies Falcon 9, ships as a free-build template. Load it, run it, watch it hold its operating point. Then start turning knobs: the feed pressures, the startup schedule, the bell, the cooling, and find out what breaks and why.
If you'd rather earn it, Mission 13 opens Chapter V, Legendary GG Engines, and walks you through building it yourself: you place the Turbopump and wire up its slots, wake the pre-placed Thrust Chamber with its headline geometry, author the startup, then tune the gas generator until everything holds at steady state. No partial credit: get the sequence right, and the engine runs nominal. If you stall, every objective can name the specific thing that's wrong and walk you to it.
The startup is the good part. Real engines start with many valves on one schedule, and you author that schedule. Each valve has its own level, width, and delay. A master ramps from 0 to 1, and the valves fire as they cross their thresholds. The editor shows you the sequence before you run, with stacked preview charts where the threshold is drawn as a shaded band, so you can scrub the master and watch each follower fire in order. Get the sequence right, and the engine runs nominal. And it isn't Merlin-only. Any engine you build can start that way.
This is inspiration, not a digital twin, and the mission says so out loud. SpaceX has never published Merlin's internal geometry, so every dimension is derived or back-solved. The real Merlin runs about 214 kg/s of LOX and 91 of RP-1 at roughly 9.7 MPa in the chamber; the debrief shows you how close yours landed, down to a shaft speed within about 5% of the real turbopump's. The physics is real. The geometry is inspired.
Undo and redo
Undo works across the editor now, on Windows, macOS, and Linux. Delete the wrong component in the middle of a tune, and one keystroke puts it back, along with its links, its chart traces, and its samplers. One gesture is one step, so undoing a five-node drag takes one press. Delete something and the notification carries an Undo button right there. It's in the toolbar, the right-click menu, the Edit menu, and the shortcut your platform expects.
The numbers underneath
The physics underlying all of this is written up properly now: a paper on the coupled engine model that powers Meco, the system solve, the nozzle flow field, and how the two agree, with a Merlin 1D-class engine as the case study. It's public, with a DOI, so you can check my work:
https://doi.org/10.5281/zenodo.21864769
In the same spirit, the Isp on the thrust card is engine Isp now; the number real datasheets quote reflects the propellant the gas generator burns, not just what the main nozzle sees. Nozzle-flow Isp is still there, labeled as such.
Those conventions are checkable. The built-in Vulcain 1 sits on its published numbers: 1140 kN of vacuum thrust and an engine Isp within 1% of the published 431 seconds.
Pumps report available NPSH next to their curves and flag cavitation when the inlet pressure drops below the propellant's own vapor pressure, not a generic threshold. Pump slip comes from the Wiesner correlation, and pump losses split into the head you actually keep and what the shaft eats. Shafts and gear meshes cost you real torque. Thrust pays for the drag along the nozzle wall.
That last group is a physics change, not a display change, so it comes with a warning: pumps now produce roughly 9 to 13% less head and torque than they did in 0.5. A saved engine that used to close may need a retune. If yours suddenly runs lean, that's where to look first.
Polish for the editor
Right-click or long-press opens the menus; a single click just selects
Error messages name the control you can see, not an internal path
The node graph comes in light now; pick your theme in Settings
Copy and paste prompts lead with the keyboard shortcut for your platform
Boundary node types explain themselves with tooltips
The control parameter preview stopped stealing your scroll position
Chart legends hold still while you hover
Also in 0.6
Missions keep your place. The hub is organized into chapters, with a one-line "what you learn here" under each mission. Missions open on their brief; finishing one lands you in the next, and Continue drops you exactly where you left off.
Progress sticks. Objectives say exactly what the sim checks, experiments and renames don't cost you credit, and a save stuck partway through Mission 6 unsticks itself on load.
Music and SFX have their own volume sliders
Helium joined the cold-gas options
Every gas path in the sim runs on a new solver that copes with layouts the old one couldn't, and the fluid solver holds up at short pipe joints, where it used to give up quietly
If you play and have a minute, a Steam review goes a long way for a small project like this.
Mind the startup transient.