AI agents can redesign engineering parts with more than a thousand CAD parameters and nonlinear simulations
Authors: Alexandra Starostina, Alexander Hermann, Mahish Guru, Kartik Bali, Roland Aydin, Christian C. Cyron
Organizations: Helmholtz-Zentrum Hereon, Institute of Material Systems Modeling, Max-Planck-Straße 1, 21502 Geesthacht, Germany · Hamburg University of Technology, Institute for Artificial Intelligence and Simulation in Mechanics, Eißendorfer Str. 42, 21073 Hamburg, Germany · Saarland University, Research Group Data Driven Simulation and Analysis in Material Science, Campus, Building D3.3, 66123 Saarbrücken, Germany
Simulation-driven design of real engineering components has to meet three conditions at once. It runs on few solves, since a nonlinear finite-element simulation of a 3D part takes hours; its design space has many values, the hundreds to thousands of values of the part's CAD model; and a better design often needs a redesign, coordinated changes of the part's proportions rather than local polishing of its surface. Automated optimisers meet at most two of these conditions. Classical optimisers, Bayesian or derivative-free, need a person to cut the design down to a few parameters, and on all the values they need far more solves; industrial shape optimisers move every surface node of the mesh in one solve per cycle, but only in small steps that polish rather than redesign. We ask whether a general-purpose large language model (LLM) can meet all three by joining what it knows with what the simulation shows. GPT-6 Astra edits a magnesium stent through a CAD geometry parametrisation of about 1,650 values, with a budget of 15 simulations; every proposal is built in CAD, meshed and simulated with large-deformation plasticity and contact, and the LLM reads the full results of every solve. Within that budget it lowered the objective, a high percentile of the strains in the stent's bends, by 32%; on the same budget, the industrial shape optimiser Tosca improved the original design by a tenth and then stopped improving. An earlier GPT-6 Astra design, 25% lower in the same objective, was checked in an independent, higher-fidelity model of the full deployment with crimping and damage: it was 14% less strained than the original after the full chain, and its most damaged spot began to fail later and carried 63% less damage at full expansion. LLMs can thus join their learned knowledge with simulation data to adjust more than a thousand design values and reach valid designs with real improvements.
Figures & tables
Figure 1: Three conditions of simulation-driven design, in colour at the corners. Each family of automated optimisers meets the two conditions its side joins and gives up the third, marked in that condition’s colour; the agent of this paper meets all three.
Figure 2: Overview of the LLM-driven optimisation loop, which runs in rounds of up to three designs. Top: the agent’s input, the LLM with its knowledge, and the agent’s output, the design files it writes. Bottom: the pipeline that turns each file into a scored simulation without the LLM; orange: the parts each problem supplies ( Table 1 ). Example entries from the stent.
Stent ( Section 4 )
Plate ( Appendix C )
Crown ( Appendix D )
Agent writes
Shared settings, every crown and link, bumps on the crown edges; about 1 650 values ( Table 2 ); outlines wrapped onto a tube
The hole’s two semi-axes and up to eight bumps on its edge; up to 26 values; outline extruded
Three widths and a shift of the apex; four values, written into the solver input
Free check
Rules, CAD build, measures
Rules, CAD build, measures
Rules on the widths
Simulation
Balloon expansion and free recoil: large deformation, plasticity, contact; about two hours per solve
Tosca’s example: linear elastic, two load cases; under a second
One crown opened and released: plasticity; one to two minutes
It receives
Strain per crown, link and ring; lumen, shortening, volume
Stress along the edge of the hole, both load cases
Plastic strain along the crown, edge by edge
Full results
Every node and element of every frame, in a Python sandbox
The same
The solver’s output files
Objective and limits
90th percentile of the crown strains; lumen, shortening, volume, length, build rules
Peak von Mises stress at the hole; material area
Mean strain of the most strained 2% of the metal; opening, shortening, volume
Table 1: Input and output of the agent for each problem. The plate runs through the same tools and geometry code as the stent; the crown ran in a lighter implementation of the same protocol, without sandbox or CAD ( Appendix D ).
Figure 3: Deployment of the original design in the full-cycle model of Steglich and Besson [41] , with the parameters of the magnesium alloy ZX10: (a) as cut, (b) crimped onto the balloon, (c) expanded by the balloon and (d) after the balloon’s release. Side views at one scale, axis upright, front half only, each with the stent’s outer diameter; the crimper and the balloon are not drawn. Colour: cumulative plastic strain, each element’s largest integration-point value. Bottom: the steps each model runs, with its run time on one or two CPUs. The reduced model used in the optimisation ( Section 4.2 ) covers only (c) and (d): it leaves out crimping, and a rigid cylinder expands the as-cut stent until the cylinder is 21.6 mm across and is withdrawn.
Figure 4: The stent’s geometry parametrisation, from a file to the solid. (a) An example design file as the agent writes it (its schema tag omitted): shared settings, where crown_offset_mm is the tip narrowing d , an entry for one crown with two bumps (red) and an entry for one link (blue). (b) The cut pattern it describes, unrolled onto the plane at equal scale. Crowns are named (ring, θ ) and links (gap, θ ), where gap g lies between rings g and g+1 . (c) The solid that FreeCAD builds from the file, with the same crown and link coloured. (d) The shared settings at one crown: strut width w , tip width w−d and leg length L . (e) The two bumps of crown (2, 150∘ ): material added on the outer edge and removed from the inner edge. (f) The link in gap 1 at 120∘ , with a narrower and longer waist, shifted by 0.20 mm around the tube. Dashed in (e) and (f): the same file without its crown and link entries.
Level
Control
Range
Shared
Strut width w
0.76–1.10 mm
Tip narrowing d
0–0.40 mm
Leg length L
0.40–1.40 mm
Ring pitch, once or per gap
7.60–8.75 mm
Link waist width
0.57–1.20 mm
Link flat
0–2.0 mm
Table 2: The stent’s geometry parametrisation: the values a design file can set, and the rules every design must pass before it is simulated. Angles θ are measured around the tube, angles φ from a crown’s apex.
Figure 5: The agent’s two runs on the stent: the score Q of every design, the 90th percentile of its 60 crown strains ( Section 4.3 ). The line is the best score among the valid designs so far. (a) The pilot sessions, which developed the loop; shaded and named, the stages over which the geometry parametrisation grew at the agent’s request. (b) The clean run, with the complete parametrisation from the first call; shaded and labelled R1–R6, its six rounds ( Section 3.3 ), with the designs of each round in the order submitted. Dotted: the best score of the pilot sessions, which the clean run passes with its tenth design.
Figure 6: Geometric differences between the agent’s design and the original, as cut. This design gives all crowns of a kind the same shape. Grey: metal in both designs; blue: only in the original; red: only in the pilot design. (a) The 44 unlinked crowns: thinner struts, a larger inner radius and, from two bumps on the outer edge, a wider tip; crosses: where each design’s most damaged element sits in the full-cycle model ( Figure 8 ). (b) The 16 linked crowns: thinner struts, with two bumps on the inner edge that widen the shoulders; the link is unchanged.
Figure 7: Damage of the agent’s pilot design and the original in the full-cycle model of Steglich and Besson [41] with ZX10: both crimped, released and expanded, the two expansions from the same model set-up. The porosity of each design’s most damaged element at full expansion, followed through the expansion. Dots: where it reaches 0.3%, the onset of void coalescence; right: the values at full expansion.
Figure 8: Damage fields at full expansion, in the model of Figure 7 : (a) the original, (b) the agent’s pilot design, each element coloured by its porosity read as there, on one scale; the darkest blue is below 0.3%, the onset of void coalescence. Left: the half of the stent around its most damaged element, seen from outside; circle: that element’s crown. Right: that crown, looking into its bend onto the cut wall where the element sits; arrow: the most damaged element and its porosity.
Figure 9: Number of design values the agent set in its clean run, out of the stent’s about 1 650, design by design. Light: the values in which a design differs from the original; dark, with their number: those in which it differs from the design submitted before it, within a round its sibling. Shaded: the six rounds. Dashed: all values of the geometry parametrisation; dotted: the values set at least once over the run, nearly all of them in edge profiles. Logarithmic scale.
Figure 10: Optimisation progress on the stent: the score Q of every design, the 90th percentile of its 60 crown strains (lower is better), against simulations, all judged by the head-to-head reading of the limits ( Section 4.3 ). Each marker is one simulated design, filled if it meets every limit and hollow if it breaks one; the line is the best score among the valid designs so far, as in Figure 5 . From the original design, fifteen simulations each: the agent’s clean run, with its rounds R1–R6, and Tosca with its out-of-the-box set-up ( Section 5.4 ). The agent’s first simulation is the original design. Green: Tosca with the same set-up from the agent’s one-shot fix, a design with uniform crowns, whose own simulation is its first; one run of fifteen simulations, capped after five to free the cluster and resumed by Tosca’s own restart.
Best valid design of
Q
Most strained element
Most strained crown
Median crown
Beyond the data
Agent, clean run
−32%
−32%
−32%
−31%
0
Tosca, from the original design
−10%
+5%
−3%
−11%
0
Tosca, from the agent’s one-shot fix
−20%
−17%
−21%
−20%
0
Table 3: Each method’s best valid design on the stent on five measures: the change against the original design, solved in the method’s own pipeline (lower is better), and the number of elements strained beyond the end of the material data. For Tosca’s run from the agent’s one-shot fix, the change includes the fix’s own gain.
Crown
Plate
Stent
Design values
4
Up to 26
About 1 650
Proposed designs
12
14
14
Ratio of design values to proposed designs
Below one
About two
Over a hundred
Classical optimisers
On all four values: Bayesian optimisation 8.6–15.4% (median 10.2%), COBYLA 0–5.8%, random search nothing
Bayesian optimisation on the two semi-axes a person picks: 27%, an ellipse
Not run as a comparison
Shape optimiser (Tosca)
–
As shipped, on its own mesh: 36%
From the original: 10%, then drifted; from the agent’s fix: a further 14% (12% on a fresh mesh)
LLMs
Complete sessions 10.1–22.3%; the best of four LLMs 19–22%; one LLM ahead of the median seed in two of three sessions
27–47%; 9 of 12 runs ahead of Tosca
32% (clean run), 26% (pilot sessions)
Table 4: Three problems with four, 26 and about 1 650 values and similar budgets: the methods that can take part, and their results. Percentages: the best valid design’s objective below the start design’s, within the budget. Details in Appendices C and D .
Appendix figures & tables7 assets
Supplementary material from the paper’s appendix.
Appendix
Run
LLM
Budget and tools
Caps
Cost
Stent: pilot sessions ( Section 5.1 )
Single call
GPT-6 Astra, high
One reply, no tools, whose design is simulated; also three one-call variants ( Section 5.3 )
6 per call, 15 in all
2.86
Four shared settings
GPT-6 Astra, high
Three proposals, one call each, each after the results of the one before
6 per call, 15 in all
1.69
Crown by crown
GPT-6 Astra, max
Three simulations, one design at a time; 40 checks; tools as in Table 6
150; 12 per call; 150 turns
7.31
Edge profiles
GPT-6 Astra, max
Nine simulations in rounds of up to three; 60 checks
150; 12 per call; 150 turns
17.47
Stent: the finished loop
Appendix
Table 5: The agent’s runs and the settings of their protocols; changes during a run are given in the text. Reasoning effort as sent to OpenRouter; the crown’s LLMs ran at temperature 0.2 with their default reasoning. Budget: simulations of the agent’s designs, beyond the start design’s own; checks: free design checks. Caps and costs in US$ of LLM calls, per session unless stated; a turn is one LLM call. Costs as billed, all sessions of a row together; the single call’s include its three variants, and the crown’s its set-up sessions, which were billed together.
Tool
Description
Arguments
python
“Run one Python program in the sandbox and return what it prints. /data and /astra_lib are read-only; /work is writable and persists between runs. No network; up to 10 minutes per run.”
code : “Python source”
view_image
“Show a PNG or JPEG file saved under /work; it appears in the next message.”
path : “e.g. /work/outline.png”
check_design
“Free checks and CAD build of a design file under /work (schema astra-direct-002/design-1). Reports pass or fail, the failures, volume, length, link flank radii, profiles, what changed relative to D0 and whether it was already simulated. Up to 60 checks.”
path : “e.g. /work/d8.json”
submit_designs
“Simulate one to 3 designs at once, in parallel (about 2.5 to 3 hours per round; the conversation waits), and return all their results. Each must pass check_design and must not repeat a simulated design. 14 simulations in total.”
designs : one to three, each with path (“design file under /work”), reasons (“why this design”) and predictions of Q , inner_d_after_recoil_mm , shortening_mm , max_crown_strain and hottest_crowns (“the three hottest crowns, hottest first, as ‘ring@theta’ ”)
finish
“End the session.”
best_design (“id of your best design, as named in the results”), summary and wanted_but_not_expressible (“changes you wanted but could not express with these controls”)
Appendix
Table 6: The clean run’s tools, with the descriptions the LLM received, verbatim. The plate’s sessions had the same tools, with the descriptions and predictions of their problem.
Method
Design space
Best per run (MPa)
Below the circle
Claude Fable 5.2, three runs
Up to 26 values
93.0, 92.7, 91.8
46–47%
GPT-6 Astra, three runs
Up to 26 values
101.5, 102.6, 94.7
40–45%
Kimi K3.1 Pro, three runs
Up to 26 values
103.7, 110.2, 115.0
33–40%
Mistral Large 4, three runs
Up to 26 values
125.6, 107.4, 105.4
27–39%
Tosca, as shipped
Its mesh: 17 edge nodes
109.4
36%
Bayesian optimisation
Two semi-axes
125.9
27%
Appendix
Table 7: The plate with a hole: the best valid design within 15 solves, the start’s included, on the judge’s fine mesh (the circle: 171.6 MPa). The agents changed the hole’s two semi-axes and up to eight bumps on its edge, up to 26 values.
Figure 11: The plate on Tosca’s own coarse model, with 16 elements along the quarter hole. (a) Tosca’s best design within 15 simulations, as its own mesh with its own loads. (b) The same design on the coarse model the agents were given: the same 17 edge nodes, with the shipped mesh’s straight lines inside where Tosca smooths them, which changes the stress at the hole by at most 1.1%. (c) The best agent design on that model, by Claude Fable 5.2. Colour: von Mises stress, the larger of the two load cases; circle: the edge node with the highest stress, where each value is read; dashed: the original hole. (d) The hole shapes: Tosca’s 17 edge nodes and the curve the judge reads through them, and each LLM’s best design from its runs on the coarse model, with the judge’s value in brackets. (e) The best valid design so far against simulations for every run on the coarse model, judged on the fine mesh, and Tosca’s reading on its own mesh.
Figure 12: The reduced crown problem. (a) The model: one unit of a zigzag ring, unrolled into the plane, from the middle of one crown down its strut to the middle of the opposite crown, with its neighbours mirrored in grey. The four values set the width at the apex, wa , 45∘ from it, w45 , and along the strut, ws , and an outward shift s of the apex; both crowns take the same values. Dash-dotted: the symmetry planes, one held, the other moved by 0.5 mm and released. (b)–(d) The plastic strain after release in the upper crown, the box in (a); the lower crown mirrors it. (b) The start design. (c) The best design of Bayesian optimisation over its five seeds. (d) The best LLM design, by Mistral Large 4; the best designs of GPT-6 Astra, Kimi K3.1 Pro and DeepSeek V4.1 Flash are alike, with a wider apex and shoulder and a thinner strut. Outlined: the most strained 2% of the metal, whose mean strain is the objective; below each panel, its four values. Each design solved again from its own input, which reproduces its recorded result exactly.
Main reading
Strict reading
Method
Run
Cap
Simulations
Study
Finer
Study
Finer
Cost
COBYLA
step 0.1
–
12
0.0
0.0
0.0
0.0
–
step 0.2
–
12
5.8
5.6
0.0
0.0
–
step 0.4
–
12
4.8
0.4
0.0
0.0
–
Random search
seed 0
–
12 (1 failed)
0.0
0.0
0.0
0.0
–
seed 1
–
12 (1 failed)
0.0
0.0
0.0
0.0
–
Appendix
Table 8: Every crown session and classical baseline, each from the same start design with twelve simulations: the peak strain of the best feasible design, in percent below the start design’s, under the main reading (the tolerances fixed in advance) and the strict reading (every limit exact), on the study mesh and on a mesh 1.5 times finer, where the design is solved again and judged against the start design on that mesh; 0.0 means that no feasible design beat the start. Cap and cost: US$ per session at list prices, helper agents included; “cap”: ended by its cost cap; “no answer”: ended by empty replies. For Kimi and DeepSeek, the second group of three ran with the larger cap.
Figure 13: The reduced crown problem: the best feasible design so far against simulations, main reading, study mesh. (a) The classical optimisers; random search and COBYLA with step size 0.1 found no feasible improvement. (b) Every LLM session under the final caps for its LLM, except the one that left the study, over the median and range of Bayesian optimisation.
Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA · Department of Mechanical Engineering, Department of Engineering and Public Policy, Dept. of Civil & Environmental Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA