A tube of springs
Fold a strip of paper into a ring of triangles and close it, and you get a tube that shortens as it twists. Yasuda et al. call it triangulated cylindrical origami. Paper panels warp and wear as the tube folds, so they keep only the creases: n valley creases of length a and n mountain creases of length b, each a linear spring of stiffness k, between two n-sided polygons on a circle of radius R.
The cell has two coordinates: the compression u from the starting height h0, and the extra twist ϕ of the top plate from the starting angle θ0. Each crease is the hypotenuse of a right triangle whose legs are the height and a chord of the circle, which gives Eq. 1 of the paper’s Methods. The stored energy is the sum of the spring energies.
The second stress-free shape
A shape with no stored energy needs a=a0 and b=b0 at once. Subtracting the two crease formulas removes the height: b2−a2=4R2sin(α+β)sin(π/n). This must keep its starting value, so sin(α+β) must too. Besides the start, the only way is α+β=π−θ0, a total twist of π−θ0 and an extra twist ϕ2=π−2θ0.
Putting this twist back into either crease gives the height H2. The shape exists only if H22>0. Zero is the lowest energy the springs can have, so the shape is then a second minimum: the cell is bistable, and geometry alone holds it shut.
Three kinds of cell
Yasuda et al. build cells with R=90 mm and n=5. With (h0,θ0)=(150 mm,40∘) their energy along the valley has two minima; with (90 mm,46∘) it only rises. The formula agrees: the first has H2=49.2 mm, the second H22=−10108 mm2, so no second shape.
A third outcome sits outside the formula. With no second shape the energy may still fall all the way down the valley, and then the cell collapses until its plates meet. The springs alone would let the plates pass through each other into a mirror cell, which no real tube can do. The Lab therefore adds an assumption of its own, outside the paper’s truss: a stiff contact that keeps the plates 0.1 R, 9 mm, apart.
A start with no stiffness
Hold the twist free and push. The cell’s first resistance is the axial stiffness at the start, the ratio of the energy’s Hessian determinant to its twist entry. Writing the crease derivatives out, the numerator is proportional to (sin2α0−sin2β0)2, which vanishes when α0+β0=π/2, that is at θ0=π/2, whatever h0, R, n or k. Yasuda et al. state that this zero-stiffness mode does not depend on k or h0 and give the proof in their supplement.
The second shape explains it. At θ0=90∘, ϕ2=0 and H2=h0: the two stress-free shapes coincide, and the energy near the start grows as the fourth power of the compression. At 92∘ they are 3.2 mm and 4∘ apart, and the stiffness is 6.86×10−4nk, 0.112 % of the monostable cell’s 0.610 nk and 0.080 % of the bistable cell’s 0.853 nk. The paper reports 0.26 % and 0.18 % for its prototypes, which are built with a calibration angle of 9.7∘; both pairs of numbers are tiny.
Folding until a sheet meets itself
Diamant and Witten (2011) meet the same ending in a different sheet. Their film floating on a liquid buckles into a fold that deepens as it is pushed. Its exact shape keeps working past the point where the fold touches itself, at a pressure of about 0.040 and an end displacement of about 5.6 in their units, but beyond it the solution passes through itself and is unphysical.
In both cases the elastic energy alone does not know that a sheet cannot cross itself. A model has to add that rule, and where the motion ends then depends on it.
Worked example
The cell that clicks shut
Take h0=150 mm, θ0=40∘, n=5, R=90 mm. Then π/2n=18∘, and the creases start at a0=(1502+32400sin22∘)1/2=150.13 mm and b0=(1502+32400sin238∘)1/2=186.50 mm.
The bracket in H22 is sin22∘−cos238∘=0.00122−0.62096=−0.61974, so H22=22500−20080=2420 mm2, H2=49.2 mm, and the top plate turns a further 180∘−80∘=100∘. At the total twist of 140∘, 32400sin252∘+2420=22539=a02 and 32400sin288∘+2420=34781=b02. In the Lab the gripper lets go at a compression of 0.233h0=34.9 mm, where holding would need a pull, and the cell settles at 49.2 mm and 100∘.