OpenSees ShellMITC4 smoke test -- clamped square plate
  a        = 1000.0 mm
  t        = 76.2 mm
  E        = 70.8 MPa
  nu       = 0.3
  q        = 1000.0 Pa  (= 1.000e-03 MPa)
  D        = E*t^3 / (12(1-nu^2)) = 2.869e+06 N*mm
  w_theory = 0.00126 * q * a^4 / D = 0.4392 mm

Building mesh: 10x10 quads -> 121 nodes, 100 quads
  boundary nodes (clamped, all 6 DOF): 40 of 121

Solving with OpenSees ShellMITC4 quad wrapper...
  log_tail: OpenSees analyze() returned 0 (converged) using element type 'ShellMITC4'; n_nodes=121, n_quads=100; stress recovered via displacement field (eleResponse returned zeros in this build)

Results:
  centre node index : 60  at [500.0, 500.0, 0.0]
  w_centre (FEA)    : 0.4866 mm
  w_centre (theory) : 0.4392 mm
  ratio FEA/theory  : 1.1079
  |error|           : 10.79 %

  max |disp| (any node): 0.4866 mm at (500.0, 500.0, 0.0)
  max von-Mises        : 0.030 MPa
  p99 von-Mises        : 0.030 MPa

Pass criterion: ratio FEA/Kirchhoff in [1.00, 1.20] (Mindlin shear band)  ->  PASS

Verdict: ShellMITC4 ratio's 1.108 against Kirchhoff on the clamped plate, in the expected Mindlin-Reissner shear-deformation band. Formulation is wired correctly on flat geometry. Next step: compare against CCX S3 on the curved dome — that's the test that motivated this whole upgrade.