跳到正文
DEV Community· NovaSolver·· 12 天前AI 評分38

旋轉圓盤中心開 1 mm 小孔,孔邊應力加倍:以有限元素法核對教科書結果

A 1 mm hole doubles the stress in a spinning disk: checking the textbook result with FEM

AI 導讀

NovaSolver Studio 的結構求解器以有限元素法驗證,旋轉圓盤中心即使只有 1 mm 半徑小孔,孔邊環向應力仍約為實心圓盤中心應力的兩倍。測試使用直徑 120 mm、厚 1.2 mm 的聚碳酸酯圓盤,以 10,000 rpm 旋轉;六種模型的應力結果與公式相差均在 0.2% 以內。小孔會改變高應力區域延伸的距離。

正文

Hoop stress against radius in a spinning 120 mm disk, solid and with holes of 1, 7.5 and 15 mm radius: FEM points on the formula curves

Spin a solid disk and the highest stress sits at the centre. Drill a hole there, even a tiny one, and the stress at the edge of the hole is about twice that. It does not shrink toward the solid-disk value as the hole gets smaller.

I checked this in NovaSolver Studio's structural solver with a CD-sized polycarbonate disk (120 mm across, 1.2 mm thick) spinning at 10,000 rpm. I solved a solid disk and five centre holes from 1 mm to 15 mm radius, and compared all six with the rotating-disk formulas.

The formulas

For a thin disk of outer radius R, density ρ and Poisson's ratio ν spinning at angular speed ω, the plane-stress solution gives the hoop (circumferential) stress. For a solid disk the peak is at the centre:

σθ(0)=3+ν8 ρ ω2R2

With a central hole of radius a, the hoop stress at radius r is

σθ(r)=3+ν8 ρ ω2(a2+R2+a2R2r2−1+3ν3+ν r2)

and its peak, at the edge of the hole (r = a), is

σθ(a)=3+ν4 ρ ω2(R2+1−ν3+ν a2)

Let a go to zero and this becomes (3+ν)/4 · ρω²R², exactly twice the solid-disk centre value. A pinhole does not bring you back to the solid disk. It is the same factor of 2 you get for a small hole in a plate under equal biaxial tension, which is what the centre of a spinning disk is.

import math

E, nu, rho = 2.4e9, 0.37, 1200.0        # polycarbonate, assumed values
R, rpm = 0.060, 10000
w = rpm * 2 * math.pi / 60

solid = (3 + nu) / 8 * rho * w**2 * R**2
print(f"solid disk, centre: {solid/1e6:.3f} MPa")
for a_mm in (1, 2, 4, 7.5, 15):
    a = a_mm / 1000
    edge = (3 + nu) / 4 * rho * w**2 * (R**2 + (1 - nu) / (3 + nu) * a**2)
    print(f"hole radius {a_mm:4} mm: {edge/1e6:.3f} MPa  ({edge/solid:.2f} x solid)")
# solid disk, centre: 1.996 MPa
# hole radius    1 mm: 3.991 MPa  (2.00 x solid)
# hole radius    2 mm: 3.992 MPa  (2.00 x solid)
# hole radius    4 mm: 3.995 MPa  (2.00 x solid)
# hole radius  7.5 mm: 4.003 MPa  (2.01 x solid)
# hole radius   15 mm: 4.038 MPa  (2.02 x solid)

For geometrically similar disks, all of these scale with ω²R². Double the speed and every stress here goes up four times.

The setup

Item Value
Shape Disk, outer radius 60 mm, thickness 1.2 mm; solid, or a centre hole of radius 1 / 2 / 4 / 7.5 / 15 mm (7.5 mm is the CD hole)
Material Polycarbonate, assumed E = 2.4 GPa, ν = 0.37, ρ = 1200 kg/m³
Load Rotation about the disk axis at 10,000 rpm
Model One quarter of one half-thickness, with symmetry on the three cut planes

Nothing holds the disk. The three symmetry planes are enough to stop it moving as a rigid body, and the centrifugal load does the rest.

Solving it over MCP

The setup and solve went through Studio's MCP tools, which an AI client can call. Here is an excerpt from the 7.5 mm hole case, with Studio's replies after each arrow, translated from the Japanese session. These runs used a local development build after v0.7.6 (2026-11-10).

import_geometry  disk_a7p5.step (structural)
→ Imported disk_a7p5.step as c_20261110_102900_a5d2 (rev 1): 6 faces, ...
  f1 [free] Cylinder, area 56.55 mm2, centroid [38.2, 38.2, 0.3] mm
  f2 [free] Plane normal [0.0, 0.0, 1.0], area 2783.25 mm2, centroid [25.82, 25.82, 0.6] mm
  f3 [free] Plane normal [-1.0, 0.0, 0.0], area 31.5 mm2, centroid [-0.0, 33.75, 0.3] mm
  f4 [free] Plane normal [0.0, 0.0, -1.0], area 2783.25 mm2, centroid [25.82, 25.82, 0.0] mm
  f5 [free] Plane normal [-0.0, -1.0, 0.0], area 31.5 mm2, centroid [33.75, 0.0, 0.3] mm
  f6 [free] Cylinder, area 7.07 mm2, centroid [4.77, 4.77, 0.3] mm

update_case  f3, f4, f5 = symmetry / E 2.4 GPa, nu 0.37, rho 1200 kg/m3
             rotation 10000 rpm about z through [0, 0, 0] / mesh.refine.f6 = 0.9375 mm

check_convergence  (re-solves on coarse, normal and fine meshes)

get_results
→ ran=yes / converged=yes / mesh convergence (GCI) passed
  Max von Mises stress: 4.004 MPa

verify_case
→ Mesh convergence (max displacement, GCI): finest mesh 0.020604 mm, Richardson extrapolation 0.020604 mm, GCI 1.7e-05% (tolerance 5.0%) -> PASS

probe_results  points along the y ≈ 0 line at z = 0.3 mm, from 0.001 mm inside the hole edge to r = 60 mm

The symmetry planes (f3 = the x = 0 cut, f5 = the y = 0 cut, f4 = the mid-plane) and the hole (f6, the small cylinder) were picked out from the face types, orientations and positions. On the y ≈ 0 line the hoop direction is y, so the hoop stress is the probe's syy. The model covers z = 0 to 0.6 mm (half the thickness, with z = 0 as the mid-plane of the disk), and the profile points sit at z = 0.3 mm. The "hole edge" values below are sampled 0.001 mm inside the hole surface. In the formulas the stress does not vary through the thickness, and the Studio values at z = 0, 0.3 and 0.6 mm differ by less than 0.1%.

mesh.refine.f6 sets the mesh size on the hole face, here one eighth of the hole radius, so the steep rise near the hole is resolved. With that setting, the 2 mm and 15 mm cases barely changed in element count between two of the three mesh levels, and Studio warned that it could not judge how reliable the answer was. I re-solved those two with one tenth of the hole radius, and all six cases then passed the mesh convergence check.

Results

Case Studio (sampled) Formula Diff. vs solid disk
Solid, centre 1.996 MPa 1.996 MPa <0.01% 1
Hole 1 mm, hole edge 3.986 MPa 3.991 MPa −0.14% 2.00
Hole 2 mm, hole edge 3.989 MPa 3.992 MPa −0.07% 2.00
Hole 4 mm, hole edge 3.990 MPa 3.995 MPa −0.11% 2.00
Hole 7.5 mm, hole edge 4.003 MPa 4.003 MPa −0.01% 2.01
Hole 15 mm, hole edge 4.038 MPa 4.038 MPa −0.01% 2.02

In all six cases the hoop stress at the sampled point matched the formula to within 0.2%. Part of the small negative difference with holes comes from sampling 0.001 mm inside the edge, where the stress has already dropped a little (about 0.1% for the 1 mm hole, by the formula). The 1 mm hole has 1/225 the area of the 15 mm one, yet its hole-edge stress is only 1.3% lower.

What the hole size does change is how far the raised stress reaches. In the first figure, the 1 mm hole's curve is back within 5% of the solid-disk curve from about r = 5 mm outward. The 15 mm hole lifts the stress across the whole disk.

Inside the disk

Von Mises stress in a quarter of the disk with a 7.5 mm radius hole, brightest at the hole edge

This is the von Mises stress in the quarter model with the 7.5 mm hole. The bright ring is the hole edge. The stress falls off outward and is lowest at the rim.

In the plane-stress solution, the radial stress is zero on the hole edge, because that surface is free, so von Mises there equals the hoop stress. At the centre of the solid disk, hoop and radial stress are equal and von Mises again equals the hoop value. In the 3D model von Mises is approximately equal to the hoop stress at both places, so the "2x" carries over to von Mises.

Takeaways

  • For a CD-sized polycarbonate disk at 10,000 rpm, Studio's hoop stress at the hole edge (or solid-disk centre) matched the rotating-disk formulas to within 0.2% in six cases
  • Any centre hole, even 1 mm in radius, about doubles the peak stress compared with a solid disk
  • The hole size mostly changes how far the raised stress spreads, not the peak; all stresses scale with rpm²

NovaSolver Studio is free and open source (GPL-3.0). It solves structural, thermal and flow problems in one app and can be driven from MCP clients such as Claude Desktop.

來源:DEV Community · dev.to