ASSAYFORGE logo
Fluidraft Module 002
Generative & Inverse Design

State the target. Search every design you didn't have time to try.

You know what the chip must do — equal flow at every outlet, a target flow through one channel, a target droplet size. ASSAYFORGE searches the parameters you allow it to change, solves every single candidate with Fluidraft's deterministic solver, rejects anything that violates your hard constraints, and hands you a ranked list. Nothing touches your design until you accept a candidate — and acceptance is one undoable step.

ASSAYFORGE is a paid module inside Fluidraft — same project, same canvas, same objects. In the browser demo it appears in its locked preview; in an unlicensed Fluidraft install it runs fully, watermarked, so you can evaluate it before buying.

ASSAYFORGE inside Fluidraft: a study with ranked, solver-verified candidates next to the live CAD canvas

A study is a contract, not a wish

Every study states exactly four things — and the engine can do nothing outside them. The editors render from the same stated vocabulary the engine scores, so what you can ask for and what gets optimised can never drift apart.

Variables

What may change — everything else stays locked

An allow-list of parameters with bounds: this channel's width, 60–140 µm. Bounds are validated against the real project and against each component's own physical limits — the optimiser is never allowed to search designs the component itself refuses.

gap_serp.width · 60 – 140 µm · ✓ search space valid
Objective

What better means

Balance the outlet flows, hit a target flow, minimise driving pressure — or hit a target droplet size, scored through the DROPLET LAB module's validated model. Every objective in the vocabulary names the solver quantity it is evaluated against: nothing is promised that the solver cannot verify.

Hard constraints

What may never happen

Max driving pressure, minimum channel width (your fab's limit, via the FABRIX bridge), maximum Reynolds. Hard constraints are never violated and never traded off — a candidate that breaks one is rejected outright, kept visible with the violated limit next to the actual value, and can never outrank a feasible design.

Budget

Bounded, deterministic, reproducible

A deterministic grid sized to your evaluation budget (hard cap 500). No randomness, no magic: the same study on the same project produces byte-identical candidates, every time — proven by test. The full provenance (engine, grid, seed, solver authority) is stored with the results in your project file.

TARGETSTUDY · variables + objective + constraints ENGINE · solver verifies every candidate RANKED CANDIDATES SANDBOX · compare ACCEPT · undoable

The optimiser proposes. The solver disposes.

The gravest risk of an optimisation module in a scientific tool is that it becomes a second source of physics. ASSAYFORGE is built so that it structurally cannot:

Zero physics of its own. Every candidate is your full project with only the allowed parameters changed, solved by the exact same code path as the SIMULATE button. A test re-solves a candidate by hand and matches the engine's stored numbers to one part in a million.
Validated on a known optimum. The reference case is a four-outlet flow balancer whose best design is known from symmetry. The engine must find exactly that design — ranked first, 0.00% outlet variation — or the release does not ship. A demo video is not validation; a found optimum is.
Your design is yours. Candidates are proposals. The sandbox shows the current design and the candidate side by side — both actually solved — and accepting applies the change as one normal, undoable edit that correctly marks your simulation results stale. One Ctrl+Z reverts everything.

In the workbench

A study in the ASSAYFORGE panel: bounded variable, valid search space verdict
The study editor: an allow-listed variable with bounds, validated against the live project — “✓ Search space valid. Everything not listed stays locked.”
The candidate sandbox comparing the current design against a candidate, both solved
The sandbox: current design vs candidate, both solved by the core solver. On the reference case: 40.5% outlet variation today → 0.00% in the candidate.

Three modules, one search

ASSAYFORGE composes with the other Fluidraft modules through the module registry — never by depending on them. With FABRIX licensed, the active manufacturing process's real limits constrain the search. With DROPLET LAB licensed, “hit a target droplet size” becomes an objective, scored by its validated model — and a candidate the droplet model refuses to predict is marked infeasible with the model's own explanation, never ranked on an invented number. Without either module, ASSAYFORGE works fully on its solver-based objectives.

Activation is a file, not a phone call home

ASSAYFORGE is switched on by your Fluidraft licence — one signed, offline file carrying Core and your modules together. Install it from the licence chip and the module unlocks in place. No online activation, no account, no telemetry.

ASSAYFORGE is a subscription add-on to your Fluidraft plan — $49/month, or $490/year with two months free — and is carried by the same signed licence file. Letting it lapse never breaks Fluidraft — or your saved studies, which stay untouched in the project file and come back the moment you renew. The full suite (Core + all three modules) is $189/month or $1,890/year.

$49/mo
or $490/year — two months free · founding price · academic 50% off

Subscribe yearly →or monthly

Licence file by e-mail after checkout · academic 50%: write from your institution

  • Sold as an add-on entitlement to a Fluidraft licence — one consolidated licence file, updated when you buy
  • Independent subscription: renew Core and ASSAYFORGE together or separately
  • Try before buying: in evaluation mode ASSAYFORGE runs fully, watermarked like the rest of the product

Base Fluidraft never depends on ASSAYFORGE. A project that carries studies opens unchanged on an unlicensed install — module data is preserved, never deleted.