←Back to Home SYSTEM04 / 05 中文
SYSTEM 04 · FORMULA R&D

Before the film blows,
know whether it can.

Blowability prediction and data platform for bamboo film-blowing formulas — shopping bags / garbage bags / courier bags / food bags / mulch film

12Explainable decision rules
9Formula indicators
5Product categories
01 · FOUR OLD PROBLEMS

Not for lack of effort — the answer lives in one person's head.

How much bamboo powder, how many parts of compatibilizer, moisture held at what level — formulas scattered across Excel, WeChat and notebooks.

Problem 01Trials cost a fortune

One pilot run — startup, materials, labor — runs to tens of thousands. Plans are guesses; you learn it doesn't blow only after it breaks.

Problem 02Experience walks away

When the veteran retires, ten years of hands-on feel walks out the door with them.

Problem 03Same hole, twice

Why the last batch broke and how it got fixed — nobody remembers.

Problem 04Customers get no answers

"Can this formula be made?" only gets "let's try and see" — quotes and lead times hang in the air.

"Fire up a batch and see" — the most expensive sentence in the industry.
02 · A THREE-STEP LOOP

Predict, trial, and get sharper with every run.

The platform doesn't make the call. It documents why it judged so, and collects what the trial found.

STEP 1 · PREDICTEnter the formula, get a verdict
STEP 2 · TRIALRun a batch, log the results
STEP 3 · LEARNData flows back, the model sharpens
Verdict

Every basis traceable

Which rule fired, how many points, why — process engineers can challenge it and recalibrate.

Field logs

Two minutes on a phone

Record trials standing next to the machine: tensile, elongation, haze, thickness — straight into the library.

Feedback

Deviations become calibration samples

Where prediction and reality disagree gets flagged automatically; when data is sufficient, retraining is suggested.

The loop works not because predictions are perfect, but because every deviation becomes the basis for the next run.
03 · THREE-GRADE VERDICT

Every grade tells you what to do next.

GradeScoreMeaningNext step
Blowable≥ 80Wide process window, key parameters in the safe range, low bubble-stability risk.Schedule a pilot; validate at current parameters.
At risk55 – 79Blowable, but one or two parameters sit near the edge.Adjust the flagged items first, then re-run the verdict.
Not blowable< 55A hard flaw: unbalanced ratios, excess moisture, or an incompatible system.Don't fire. Fix the flagged rules first.
Without a verdict, there's no confidence to quote.
04 · HONEST LIMITS

Three things the platform can't do — and won't pretend to.

01

Data

Without trial data the model is empty. When samples fall short it says so plainly — no fabricated conclusions.

02

Responsibility

It provides the basis; it doesn't sign. Every verdict must be reviewed by a process engineer before execution.

03

Replacement

It doesn't replace pilot trials — it removes the pointless ones, saving money and machine time for formulas worth trying.

Keep the decision basis first — then talk about how smart the model is.

Turn "try and see" into
decisions on the record.

Formula import → blowability prediction → trial logging → data feedback → model retraining, one line

Enter the Formula R&D Platform
FAQ

Questions we are asked before onboarding

Answers below follow the same wording as the pages they summarise.

What problem does the Formula R&D system solve?
Whether a formula can be blown currently lives in one person's head, and trial-and-error is expensive. The system predicts blowability before the film is blown and turns every deviation into a calibration sample, so accuracy improves with use.
What is the blowability prediction based on?
A model built from historical formulas and trial records. Each judgement cites its evidence and can be traced back to a specific formula batch and its process parameters.
What can the system not do?
Three things it does not claim: it does not replace real trial data, it does not take responsibility for a formula, and it does not promise to replace imported raw materials. It turns 'let us try' into 'we have grounds'.