Fold Lab — Anyground

折01 Eight fold families

One crease grammar, eight ways to stand up.

Every ORIJO building is assembled from eight fold families, each with its own hinge, lock and way of moving. Every card opens the nearest simulator pattern, and says plainly when that pattern is only a stand-in.

F1 · I01

YOSHI vault

≈ 2.6×shell volume

Rigid-foldable one-DOF quadrilateral-panel barrel with degree-4 vertices

Folded: 2 half-shell packs
Deployed: 3.64 span x 7.28 long x 2.90 rise
Moves by: Gable portal winched along floor rails

EXPERIMENTALcalculated Simulate · Arc-Miura vault →

F2 · I06

MIURA wing (engawa)

37.9×plan area

Thick-panel Miura-ori with uniform-thickness correction

Folded: 1.00 x 0.70 x 0.60 stack
Deployed: 7.28 x 3.64 canopy at 2.6 m eave height
Moves by: Single linear actuator or hand winch pulling the far corner

COMMERCIALEXPERIMENTALcalculated Simulate · Miura-ori sheet →

F3 · I09

KRES mast

7×height

Bistable Kresling cylinder cells

Folded: diameter 1.20 x 0.90 high
Deployed: diameter 1.20 x 6.30 high
Moves by: One compressor pulse per cell or twist ring

EXPERIMENTALcalculated Simulate · Kresling tower →

F4 · I07

MUTO spine

≈ 5×pack to span

Uniformly thick degree-6 diamond vertex modules and zipper-coupled Miura tubes

Folded: 1.82 x 0.91 x 0.25 cassette
Deployed: deck or bridge 0.91 wide x 3.64-4.0 m span
Moves by: Linear actuator 6-8 kN or manual lever

EXPERIMENTALcalculated Simulate · Miura-ori sheet →

F5 · I13

RESCH roof

33–66 %fold, by season

Thick-panel Ron Resch star-tuck tessellation with seasonal partial-fold states

Folded: 2.00 x 1.90 x 0.40
Deployed: 5.46 x 1.82 roof at 33% fold or 66% fold
Moves by: One linear actuator at the central vertex

EXPERIMENTAL Simulate · Waterbomb shell →

F6 · I18

KIRI shoji skin

≈ 6×area

Lockable rigid-foldable origami-kirigami cells and single-string kirigami tiles

Folded: 0.91 x 1.82 x 0.08 panel
Deployed: gable
Moves by: One string pulled by hand or TSURU

EXPERIMENTALtarget Simulate · Kirigami facade →

F7 · I19

KASA curved-crease canopy

n/aflat → curved

Curved-crease shells approximated by self-shaping hygromorphic timber bilayers and bistable auxetic sheets

Folded: flat bundle 3.64 x 1.82 x 0.30
Deployed: single-curved canopy up to 7.28 x 3.64
Moves by: Controlled drying in transit or on site or push-into-place

EXPERIMENTALSPECULATIVE Simulate · Arc-Miura vault →

F8 · I07

HOSHI gravity dome

14×volume

Pyramidal-Based Approximated Dome thick-panel pattern deployed by gravity and wire tension

Folded: 2.00 x 2.00 x 0.60
Deployed: 3.64 x 3.64 plan x 2.60 high
Moves by: Lift crown by tripod winch or telehandler

EXPERIMENTALcalculated Simulate · Waterbomb shell →

厚02 Thick-panel rigid origami

Paper has no thickness. A wall has 80 millimetres.

Fold two 80 mm insulated panels about a line through their middle and they crash into each other at once. There are two proven ways out, and each ORIJO test article uses exactly one of them.

Fold angle 70°

aPaper, t = 0folds 180° flat

The mathematical model. Rigid-foldability is a property of this sheet first.

bThick, hinged mid-plane—

Material on the closing side has nowhere to go. Clash grows with t · sin θ.

cTachi tapered—

Hinge on the surface, edges bevelled. Folds cleanly until the bevels meet, which gives a built-in stop but a limited range. Used for the Miura wall.

dChen / You offset—

Hinge moved to the face that closes. Folds fully flat into a stepped stack, and the far face opens by t · Δθ. Used for the vault.

Sections drawn to scale: t = 80 mm temperate panel, 400 mm of each facet shown. Shu red marks material that would have to occupy the same space.

One degree of freedom

A rigid-foldable pattern moves like a linkage: one input, here a gable portal winched along the floor rails, sets every crease angle at once. There is no transition front and no moment when part of the shell is open and part is closed. Tachi (2009) showed quadrilateral-panel cylinders that do this; the YOSHI vault is one.

The gate before tooling

A crease pattern goes to the CNC only after a kinematic simulator shows zero facet strain across the full fold. Anything that needs panels to bend or snap through, such as the classic Yoshimura, is either rejected for thick panels or redesigned as a compliant part with its own snap-through test.

target 0 % facet strain, MERLIN or Origami Simulator check

One method per article

Tapered panels fold only part-way, so the Miura wall takes its folded thickness from geometry. Offset panels fold flat but need stack depth for the stepped hinge lines, so the vault packs as 2 × 3.70 × 2.30 × 1.10 m. No packing figure assumes both methods at once.

否03 The rejected vault

Why the most famous origami vault is not in our catalogue.

The Yoshimura diamond pattern is the textbook origami barrel, and it was our first sketch. Hold its springing lines at span and push it together like a bellows: its diagonals would need to shorten by about a fifth. Paper puckers quietly. Fibreglass sandwich panels crack. So we replaced it.

Rejected

Yoshimura vault吉村パターン

—strain now

Adopted · F1 YOSHI

Arc-Miura · YOSHI vault円弧三浦

—strain now

70% strain0.05 – 8 %

Peak panel strain through the fold modelled

Yoshimura vault 24.1%Arc-Miura vault 0.033%Kresling tower 9.7%Waterbomb shell 5.2%Miura-ori sheet 0.001%Kirigami facade 0.002%

Yoshimura peaks at 24.1% in a 21-step sweep (22 % at the 70 % checkpoint of the engine tests). The arc-Miura stays at 0.033%, which is solver noise. Strain is per bar, at equilibrium, in the same engine that runs the lab above.

Where it strains: the flat crease patterns

Yoshimuraworst state at 85% fold · peak 24.1%

YOSHI · arc-Miuraworst state at 15% fold · peak 0.033%

重04 M01 · KASANE hinge

The crease is the weakest seam. So it seals, stops and locks itself.

Two membranes and a gasket replace taped seams. A tight membrane spans the face the crease turns about; a slack bellows spans the face that opens; GFRP closeouts take bearing at the stop; a thermally broken latch carries the tension.

Macro cross-section of a KASANE hinge between two insulated sandwich panels (concept visual) — I02FRP skins, PIR core, GFRP closeouts, membrane and bronze latch.Concept visual · AI

V02 · Hinge breath

V02The crease breathing through its fold range.Concept video · AI

Fig. 04 · Section through one crease, 1:4Latched at design stop

1GFRP skins · outer = rain screen, inner = vapour control
2PIR core
3Pultruded GFRP edge closeout · bears at the stop
4Aerogel strip · keeps the thermal line
5Axis-face membrane · tight, zero length change
6Opening-face bellows · ≥ 1.2 · t · Δθ
7EPDM bulb gasket · second air barrier
8Over-centre latch · GFRP body, bronze cam, Hall sensor

Panel

Fold change from the stop, Δθ0°

Opening-face gap calculated

0mm

Measured on this drawing; t · Δθ = 0 mm along the face.

Bellows fitted calculated

151mm

1.2 · t · π/2 for a 90° fold change; 0 % of it is in use.

Latch tension calculated

7.1kN / latch

T = M / 0.8t at 910 mm pitch, illustrative M = 0.5 kN·m/m until the FE wind case.

Panel-centre U calculated

0.28W/m²K

λ 0.022 / t, surface resistances ignored. Polar sites need 200–250 mm.

Parts

Part	Material	Spec
Facet panel	Glass-FRP faces, PIR core (mycelium optional), aerogel strip behind edge closeout	80-120 mm temperate, 200-250 mm PIR-equivalent polar; 10-12 kg/m2 at 120 mm PIR (calculated)
Edge closeout	Pultruded GFRP channel or hardwood	15-25 mm deep bearing surface; allowable bearing stress from closeout datasheet, not from the core (PIR about 0.15-0.3 MPa)
Axis-face membrane	PVDF-coated polyester, vacuum laminated	tight across crease; target 2,000 fold cycles
Opening-face bellows	Slack PVDF-polyester loop, or EPDM/silicone sheet (>= 300% elongation) bonded to membrane	length >= 1.2 x t x dTheta (150 mm at 80 mm, 230 mm at 120 mm, 90 degree fold)
Over-centre latch	GFRP or polyamide body, bronze cam at bearing point only, stainless pin	primary tension member; T = M_crease / (0.8 t); 910 mm pitch; inside face behind vapour layer; Hall sensor
Backup gasket	EPDM bulb	compressed 30-40% at preload (target)
Obi repair kit	Membrane roll, adhesive, portable vacuum pump	300 mm patch, 30 min (target)

Maturity

EXPERIMENTAL Every ingredient exists as a prototype or lab result. The weathertight, insulated building hinge with an opening-face bellows is unbuilt, and it is the single biggest risk in ORIJO. The 12-month gate: water spray on every opening crease, q50 ≤ 1.5 m³/(h·m²) on a 2 × 2 × 2 m box, 2,000 powered fold cycles.

How it works, in 12 steps

Facet panels are glass-FRP sandwiches with a PIR core: 80-120 mm for temperate sites; polar sites need 200-250 mm PIR-equivalent (U about 0.09-0.11 W/m2K at panel centre). A mycelium core is an option only where mass and U targets allow (it adds 7-24 kg/m2 at 120 mm).
Every panel edge carries a structural closeout of pultruded GFRP or hardwood, 15-25 mm deep. This closeout, not the core, is the bearing surface; a 10-20 mm aerogel strip sits behind it to keep the thermal line continuous.
Each test article uses one thickness method. The Miura wall uses Tachi tapered panels, which have a limited fold range, so its folded thickness is taken from the fold geometry. The vault uses Chen/You offset panels, which fold flat but need stack depth for the offset hinge lines. No packing figure assumes both at once.
Rigid-foldability is checked before tooling: the crease pattern is run in a kinematic simulator (MERLIN or Origami Simulator) and must show zero facet strain through the whole fold. Patterns that need facet bending or snap-through (for example the classic Yoshimura) are not used for rigid thick panels.
At each crease the rotation axis sits on one face, and that face alternates between mountain and valley creases. A tight vacuum-laminated PVDF-coated polyester membrane spans the crease on the axis face. On the opposite face the edge gap changes by about t x dTheta, so the membrane there is a slack loop or bellows of at least 1.2 x t x dTheta (about 230 mm for t = 120 mm and a 90 degree fold change; about 150 mm for t = 80 mm), or a bonded EPDM or silicone strip with at least 300% elongation.
The inner layer is the vapour control layer and the outer layer the rain screen; at each valley the space between them drains to the sill gutter, with a heat-traced or self-draining outlet at cold sites.
The offset or taper geometry lets each crease close only to its design dihedral; faces meet and stop it, so a wall cannot over-rotate even if a winch is mis-operated.
At the design dihedral the edge closeouts bear on each other in compression. Under wind reversal the tension side of each crease moment passes through the latch, so the latch is a primary structural member: latch tension T is about M_crease / (0.8 t), and it is sized for fatigue as well as preload.
The latch is thermally broken: a GFRP or polyamide body with a bronze cam only at the bearing point, mounted on the inside face behind the vapour layer so it does not pierce the membrane.
An EPDM bulb gasket inside the membrane is compressed by the same latch preload as a second air barrier.
Each latch carries a Hall-effect position sensor and a strain gauge; the module passport records engaged or not, and TSURU verifies every latch after deployment.
A failed membrane or bellows strip is cut out over 300 mm and re-laminated with a field vacuum kit as an obi patch (target 30 minutes).

KASANE, YOSHI, JO-PORT and ASHI in 3D →

比05 Expansion ratios

How much building comes out of the box.

Ratios are measured on different bases (volume, plan area, height), so each is labelled with its basis. For comparison on the same basis as the YOSHI vault's 2.6×, a TenFold TF-20 manages about 1.9×.

ID	Family	Folded (m)	Deployed (m)	Expansion log · 1× — 10× — 100×	Maturity
F1	YOSHI vault	2 half-shell packs, each 3.70 x 2.30 x 1.10 (80 mm panels; 1.60 high for 120 mm); target envelope to be confirmed by fold simulation	3.64 span x 7.28 long x 2.90 rise (YOSHI-4); 3.64 x 10.92 x 2.90 (YOSHI-6)	≈ 2.6× shell volume about 2.6x shell only (calculated: about 48 m3 interior / 18.7 m3 of packs); about 2.2x whole room incl. floor cassettes and gables (48 / 22.2); TenFold TF-20 about 1.9x on the same basis	EXPERIMENTAL
F2	MIURA wing (engawa)	1.00 x 0.70 x 0.60 stack (one wing)	7.28 x 3.64 canopy at 2.6 m eave height (26.5 m2)	37.9× plan area 63x plan area (calculated: 26.5 m2 / (1.00 x 0.70 m) = 37.9; with 0.6 m stack the area-to-pack-volume ratio is 63 m2/m3)	COMMERCIALEXPERIMENTAL
F3	KRES mast	diameter 1.20 x 0.90 high (7 cells at 0.12 m folded + base)	diameter 1.20 x 6.30 high (7 cells at 0.90 m)	7× height 7x height (calculated: 6.30 / 0.90)	EXPERIMENTAL
F4	MUTO spine	1.82 x 0.91 x 0.25 cassette	deck or bridge 0.91 wide x 3.64-4.0 m span; or column 2.73 m	≈ 5× pack to span about 5x (calculated from fold-to-span geometry; MUTOIS reports packaging ratio >5 for deployable shelters)	EXPERIMENTAL
F5	RESCH roof	2.00 x 1.90 x 0.40 (KURA roof cap)	5.46 x 1.82 roof at 33% fold (summer) or 66% fold (winter)	33–66 % fold, by season state-variable (33-66% fold)	EXPERIMENTAL
F6	KIRI shoji skin	0.91 x 1.82 x 0.08 panel	gable, partition or louvre wall to 3.64 x 2.73	≈ 6× area about 6x area (target)	EXPERIMENTAL
F7	KASA curved-crease canopy	flat bundle 3.64 x 1.82 x 0.30	single-curved canopy up to 7.28 x 3.64 (target)	n/a flat → curved n/a: flat-to-curved shaping, not volume expansion	EXPERIMENTALSPECULATIVE
F8	HOSHI gravity dome	2.00 x 2.00 x 0.60 (one pallet)	3.64 x 3.64 plan x 2.60 high	14× volume 14x (calculated: about 34.4 m3 approximated dome volume 3.64 x 3.64 x 2.6 x 0.67 / 2.40 m3 pack); PBAD prototype reported 15x	EXPERIMENTAL

calculated from stated dimensionsdesign targetnot a volume expansion

動06 In the field

Gravity, a snap, a season, a string.

Four families that each move by a different force: a dome that falls open, a mast that snaps up cell by cell, a roof that changes its fold for snow, and a skin that opens with one pull.

F8 · V03

V03HOSHI gravity domeConcept video · AI

星F8 · HOSHI gravity dome

Lift the crown. Let go. Gravity does the rest.

A tripod winch or telehandler lifts the crown of the folded pallet; released, the pyramid facets swing down under their own weight in tens of seconds. Wire tension, contact locks and a floor-ring latch then hold the dome.

Pattern: Pyramidal-Based Approximated Dome thick-panel pattern deployed by gravity and wire tension
Folded: 2.00 x 2.00 x 0.60 (one pallet)
Deployed: 3.64 x 3.64 plan x 2.60 high
Expansion: 14x (calculated: about 34.4 m3 approximated dome volume 3.64 x 3.64 x 2.6 x 0.67 / 2.40 m3 pack); PBAD prototype reported 15x
Lock: Wire tension plus contact lock and floor-ring latch
Load: Sleeping pod; dome action into floor ring on ASHI legs

EXPERIMENTALNearest in the simulator: Waterbomb shell →

F3 · V04

V04KRES mastConcept video · AI

塔F3 · KRES mast

Seven bistable cells, raised one snap at a time.

Each Kresling cell has two stress-free states. One compressor pulse or a twist ring drives a cell through its strained middle until it snaps tall; a collar pin locks it there. The lowest two cells carry a 2,000 L food-grade water bladder.

Pattern: Bistable Kresling cylinder cells
Folded: diameter 1.20 x 0.90 high (7 cells at 0.12 m folded + base)
Deployed: diameter 1.20 x 6.30 high (7 cells at 0.90 m)
Expansion: 7x height (calculated: 6.30 / 0.90)
Lock: Bistable snap per cell plus a mechanical collar pin at each stable height
Load: Lateral wind stiffness at tower scale is unverified in literature: guyed or mast-base moment frame until tested

EXPERIMENTALNearest in the simulator: Kresling tower →

F5 · V06

V06RESCH roofConcept video · AI

屋F5 · RESCH roof

A roof that holds a different fold in summer and winter.

One linear actuator at the central vertex sets the star-tuck tessellation between 33 % fold (summer) and 66 % fold (winter); mechanical pins carry the load in each state. An ESP32 controller reads the forecast and moves it ahead of snow.

Pattern: Thick-panel Ron Resch star-tuck tessellation with seasonal partial-fold states
Folded: 2.00 x 1.90 x 0.40 (KURA roof cap)
Deployed: 5.46 x 1.82 roof at 33% fold (summer) or 66% fold (winter)
Expansion: state-variable (33-66% fold)
Lock: Actuator holds state, mechanical pins at 33% and 66%
Load: Ron Resch unit tested under ASCE 7-22 snow at 33% and 66% folded; model error 6% and 17%

EXPERIMENTALNearest in the simulator: Waterbomb shell →

Janus flip facets on bistable kirigami hinges (concept visual) — I18KIRI shoji skinConcept visual · AI

障F6 · KIRI shoji skin

A shoji skin opened by one pull of a string.

Lockable hexagonal origami-kirigami cells and single-string tiles open into gables, partitions or louvres. Hygromorph and Janus flip facets switch on their own with humidity or heat; at full deployment a contact lock holds every panel.

Pattern: Lockable rigid-foldable origami-kirigami cells and single-string kirigami tiles
Folded: 0.91 x 1.82 x 0.08 panel
Deployed: gable, partition or louvre wall to 3.64 x 2.73
Expansion: about 6x area (target)
Lock: Panel contact lock at deployed state (Nat Comm 2022)
Load: Non-structural

EXPERIMENTALNearest in the simulator: Kirigami facade →

証07 What the solver has shown

Results, not renders.

Miura vs closed form modelled

< 5%

Simulated width, length and height against the Schenk–Guest solution at 30, 70 and 100 % fold

YOSHI · arc-Miura peak strain modelled

0.033%

345 nodes · 960 bars · 888 hinges, folded flat to packed in 21 steps

Yoshimura peak strain modelled

24.1%

Springing lines held at span on rails; diagonals must shorten, so panels would have to stretch

Solver speed measured

0.134 ms

per step on a 1,521-node Miura sheet (last test run), so the whole lab runs live in the browser

Configure a building → Structure systems → Open the bare simulator →