Verification

Results you can trust, and show.

Keystone is checked against closed-form solutions, against OpenSees, an independent research-grade finite-element program, and against the published ETABS and SAFE verification problems, on every release.

≤ 10⁻¹¹

relative difference from OpenSees on frames

0.3 %

core-wall building vs. OpenSees shells

0.23 %

Scordelis–Lo roof benchmark

1,000+

automated tests per release

Benchmark	Reference	Result
3-D frames: static displacements, member forces, periods, ELF loads	OpenSees	≤ 1×10⁻¹¹ relative
P-Delta (iterative, consistent geometric stiffness)	OpenSees / exact beam-column	≤ 1×10⁻⁴
Shell element: patch test, MacNeal beam, Cook's membrane, Navier plates, Scordelis–Lo roof	Closed form / published benchmarks	all pass, roof 0.23 %
Core-wall building: periods and lateral displacements	OpenSees ShellDKGQ	within 0.3 %
Cantilever shear wall: top drift	Timoshenko beam theory	within 1 %
NBC 2020 / ASCE 7-16 load generators and combinations	Hand calculations	86 checks pass
CSA A23.3 and S16 member design	Hand calculations	153 checks pass
Steel libraries (AISC, CSA, EN, IS, BS, AS/NZS)	Source tables and exact geometry	292 checks pass

Side-by-side with ETABS. The Windows app runs your installed ETABS on the same model and reports Keystone's periods, base reactions, story shears and drifts next to ETABS's, value by value. Comparisons on beta firms' real buildings are under way and will be published here with their permission.

ETABS verification manual

The same problems ETABS is verified on.

CSI publishes the problems ETABS is verified against, with theoretical results or refined SAP2000 meshes as the reference. We rebuilt every linear one in Keystone.

Problem	Reference	ETABS	Keystone
1 · Plane frame with span loads: moments and shears	Theory	exact	exact
2 · Three-storey frame: periods and mode shapes	Theory	exact	within 0.1 %
3 · ASCE 7 equivalent static storey shears	Hand calculation	exact	within 0.01 %
4, 5, 8 · 3-D frames and braced building: periods, modes, spectrum response	Theory / published	within 0.3 %	within 0.3 %
6 · Nine-storey, ten-bay frame: eigenvalues	Published	within 0.1 %	identical to ETABS
15 · Shear walls, coupled walls, wall on columns, C, E and flanged walls: lateral displacements and twists (30 values)	SAP2000 refined mesh	stiffer in 26 of 30, by up to 22 %	19 of 30 within 1 %, 25 within 4 %, median 0.4 %*

Why ETABS and the refined mesh differ on walls. In CSI's Example 15 the ETABS models use one element per wall panel per storey, which is stiffer than a converged mesh: up to 22 % on a one-storey wall, under 8 % on six storeys. Keystone with a fine wall mesh lands on the refined result. With one element per panel, Keystone comes within about 2 % (median) of ETABS’s published values (up to 12 % on the wall-on-columns case), so a difference between the two can be traced to the mesh.

*All values use Keystone's default shell walls. The outliers are the C-shaped wall under load along its web (6 storeys: 20 % stiffer in displacement, 26 % in twist; 3 storeys: 7–8 %) and a 0.0005-radian twist published to one significant figure. CSI does not state which wall element its C-wall model used; modelled as membrane walls (no out-of-plane bending), Keystone matches it within 0.3 %, while the E-wall matches as shell walls. Problems that need nonlinear links, time history or data not in the manual (7, 9, 10–14) are not included.

SAFE verification manual

Slab design, checked the same way.

CSI's SAFE verification problems for plate analysis, punching shear, slab flexure and cracked deflection, rebuilt in Keystone on SAFE's own meshes. Thin-plate results are shown; the thick-plate versions agree within 5%.

Problem	Quantity	SAFE / hand calculation	Keystone
CSA A23.3-14 RC-PN-001 · flat plate, interior column	Punching stress / resistance / D/C	1.792 / 1.127 MPa / 1.59	1.792 / 1.127 MPa / 1.59
ACI 318-14 RC-PN-001 · flat plate, interior column	Punching stress / resistance / D/C	0.193 / 0.158 ksi / 1.22	0.193 / 0.158 ksi / 1.22
CSA A23.3-14 RC-SL-001 · one-way slab, 1 m strip	Strip moment / bottom steel	25.00 kN·m / 5.414 cm²	25.00 kN·m / 5.406 cm²
Example 1 · simply supported plate, uniform / point / line load	12 deflections and 12 strip moments	all within 0.4% of SAFE
Example 2 · plate with fixed edges, point load	Centre deflection	0.0598 in	0.0598 in
Example 3 · plate with mixed edge conditions	Free-edge deflection	0.5502 in	0.5502 in
Example 4 · plate on flexible edge beams	Centre deflection / beam moment	0.1854 in / 1,120.4 kip-in	0.1854 in / 1,120.4 kip-in
Example 5 · infinite flat plate on columns	7 deflections and 8 strip moments	all within 0.5% of SAFE
Example 6 · infinite plate on elastic subgrade	Deflection under load / moment	0.05405 in / 37.99 kip-in/in	0.05405 in / 37.99 kip-in/in
Example 7 · skew plate, free obtuse edge	Free-edge deflection / max moment	1.510 in / 12.03 kip-in/in	1.511 in / 12.34 kip-in/in^†
Example 16 · cracked slab deflection, creep and shrinkage	Short-term / long-term deflection	13.55 / 24.51 mm (hand 14.4 / 23.9)	13.40 / 23.84 mm
Flat plate, 3 × 8 m bays · statics across all strips	Midspan + mean support moment = w·l2·l1²/8	2,973.6 kN·m	2,973.6 kN·m
Beam on elastic foundation (Hetényi), point load	Deflection under the load	0.931 mm (closed form)	0.934 mm
Footing with uplift, e > B/6, compression-only soil	Maximum soil pressure	266.7 kPa (closed form)	265.6 kPa

† The thin-plate moment is singular at the obtuse corner; Keystone's value converges toward the theory (11.84) as the mesh is refined.

On every run

A hand-check report with every analysis

Beyond release testing, each analysis recomputes every load case independently of the solver (gravity totals, code base shear, wind pressures, spectrum scaling) and checks drift, deflection, uplift and wall cracking against their limits, citing the clause for each.

Your numbers

Compare it with your own results.

Bring an ETABS model and we'll help you compare.

Request beta access →