Man, it's really too late/early for this, but I suspect the problems begin around here: Angular momentum is a vector quantity, and it's almost too easy to lose track of components and frames of reference even without not only having a wrong formula for L, but a wrong formula for the magnitude of L. On a first glance, your model can be simplified to a rod spun a distance r around an axis, parallel axis theorem style, but I'm gonna take this one in pieces. Updates inbound.KE = 1/2 Iomega^2. SHIT. Frickin' moments of inertia? I = L/omega. L = M x V x R. Uhhhm, M is mass is 1kg. V is angular velocity. R is radius around the centroid - hey we got all this shit!
L = r x p (L, r, p vectors) ; p = m * v (p and v vectors) |L| = m * r * v * sin(angle between component vectors r and v)