But i’ve been trialing Rhino+Grasshopper with Karamba3D to auto-size a 7-story steel frame; early runs drop tonnage about 7% while holding story drift <= H/500 and member unity checks < 0.9 when rechecked in ETABS. The concern is fidelity — joint stiffness, second-order effects, and connection eccentricities can erase the gains if the model is too ideal. What workflows or tools are you using to optimize material without compromising robustness, and how are you validating across platforms?
I got a similar about 7% drop, but it only held “H/500” in ETABS after I gave beam-to-column ends partial rotational springs. I back-calc kθ from a quick 2D ETABS joint model and plug that into Karamba supports, then run second-order so the unity checks stay under 0.9. Watch slab/diaphragm stiffness — if you leave it too rigid in GH, drifts look rosier than reality.
If you want the 7% to stick in ETABS, turn on Karamba’s 2nd‑order theory and include about 0.3% notional lateral load in GH; without it my columns came out skinny and H/500 looked fine in GH but not in recheck. @r_underwood22, your spring approach lined up once I did that — do you also model panel‑zone shear?
Matching face‑of‑column end offsets plus short rigid links in Karamba kept my about 7% when I moved to ETABS; adding a simple panel‑zone shear spring made kL and drift line up. > connection eccentricities — I capture them with about 100–150 mm beam end offsets to the node; you’ll give back about 1% steel but the drift/unity gap goes away — are panel zones on in your ETABS model?
On my last 7‑story pass, the only way the about 7% tonnage stuck was to pattern LL in Karamba like ETABS (alternate bays on/off); otherwise drift looked fine in GH but blew ‘H/500’ in ETABS. I also toggle shear deformation for deep girders — without it, member unity looked rosy in GH and then jumped back in ETABS, . If time’s tight, at least throw in ±5% accidental torsion so the recheck doesn’t surprise you.
What kept my about 7% in the handoff was matching diaphragm stiffness — use a thin orthotropic shell for the slab in Karamba so the in‑plane stiffness tracks ETABS; with a rigid diaphragm assumption the gains tended to evaporate later. If your ETABS model is rigid‑dia, skip this. @cwilson45.