Cut scrap with scientific molding
Quality

Cut scrap with scientific molding

Financial impact

$120,000USD/ yr EBITDA

Operational impact

~200fewer scrap lb / day
medium confidence

The math

Cost of one scrap point+$120,000

5,000,000 lb × 1% × $1.50/lb × 1.6 (waste markup)

Scrap removed$120,000/ yr

1 point × $120,000

Capex

$80,000

Payback

8 months

Difficulty

3 / 5

Time to implement

~20 weeks

Scrap points removed
points
Waste markup on a scrap point
x resin
Sensors and training
$

Also uses annual resin volume (lb) 5,000,000, resin cost / lb $1.5. Edit on the dashboard.

Description

Right now each press is set from the operator's experience rather than a measured, documented process, so it runs near the edge of its process window, the band of settings that still makes a good part. A small drift in the resin or the machine tips a part to scrap, and because nothing watches the process in real time, nobody catches it until a bad part reaches the inspection box at the end of the line. Scrap runs around 5% of the resin the plant buys, and resin is the biggest cost in the building, so every point of scrap is real money.

Evidence

  • Setup sheets list only "the numbers that work," with no cavity-pressure or viscosity data behind them
  • The same handful of tools show up on the scrap report almost every week
  • Bad parts are caught at the inspection box at the end of the line, not at the press
  • Technicians adjust hold pressure and cool time by hand each shift, going on experience rather than data

Deep dive

Expand for detail

Load-bearing, card breaks if wrong

A real share of the scrap comes from an unstable process, not from worn tools or bad material.

Scientific molding removes the scrap that comes from a process running too close to the edge of its window, where a small drift in melt or pressure tips a good part into a bad one. If the real cause is a worn tool that flashes, or contaminated regrind, cavity-pressure control changes little. The whole $120K rests on this. What it looks like on the floor if this is wrong: you instrument a tool, hold the process dead steady, and the scrap barely moves because the flash was coming from the parting line all along. The cheap check is to pull the scrap codes first and sort process faults (shorts, sink, dimensional drift) from tool faults (flash, drag).

The tools are sound enough to hold a repeatable pressure signature.

A cavity-pressure trace only means something if the same tool makes the same shot twice. On a worn or poorly maintained tool the trace jumps around and there is no stable window to lock onto. That is why the mold-maintenance card is the precondition, not an optional partner. What it looks like on the floor if this is wrong: the pressure trace will not repeat shot to shot, and the technicians end up chasing noise.

Key customers will fund or accept a re-validation on a workable timeline.

Improving a validated process is worth nothing until the customer signs off on the change through PPAP. If a major automotive or medical customer drags the re-validation for months, the saving slips a year. What it looks like on the floor if this is wrong: better processes sit built but un-deployed, waiting on a customer quality engineer. The workaround is monitoring-only on validated parts and full scientific molding on new tools launched right from day one.

Supportive, card weakens but survives

The shop can hire and keep technicians who can run scientific molding.

This is a skill, not a box you install. It needs two or three technicians trained in decoupled molding and cavity-pressure work, and it needs them to stay. What it looks like on the floor if this is wrong: one person can set up the instrumented jobs, they leave, and the capability leaves with them.

The scrap was tool-driven, so the sensors change nothing

Risk

Cavity-pressure control fixes an unstable process. If the real cause is a worn tool that flashes or contaminated regrind, holding the process steady does not remove the scrap.

Early Warning

The pressure trace will not repeat shot to shot, and flash returns within a run of a freshly cleaned tool.

Mitigation

Sort the scrap codes before you start, and run the mold-maintenance card first so the tools can hold a signature.

Stake

The full $120K, plus a wasted $15K to $30K per instrumented tool.

Re-validation drag stalls the rollout

Risk

A process change on a validated part generally means notifying the customer and, unless they waive it, PPAP sign-off before the improved parts can ship. If the changes pile up waiting for a customer quality engineer, the improved processes sit un-deployed for months.

Early Warning

Better processes are built and proven internally but still running the old approved set-points in production.

Mitigation

Capture the containment value with monitoring-only, batch the changes so you pay the PPAP toll once per part, and piggyback on any change the customer is already funding.

Stake

Six to twelve months of delayed benefit on the affected parts.

The trained technicians leave

Risk

Scientific molding is a skill held by people. If the one technician who can run instrumented jobs walks, the capability evaporates.

Early Warning

Only one person can set up the instrumented tools, and setups are not written down.

Mitigation

Train a bench of three, document machine-independent setup sheets, and cross-train so no single departure sinks it.

Stake

The whole $120K if the capability is lost.

Alarms get bypassed under production pressure

Risk

When the floor is behind, technicians silence the containment alarms to keep the press running, and bad parts flow again.

Early Warning

Rising nuisance-alarm counts, and operators overriding the sort gate.

Mitigation

Tune the alarms so they are trustworthy, and tie a daily supervisor check to alarm-override counts.

Stake

The scrap gain quietly erodes back toward the starting point, plus the risk of a quality escape to a customer.

A. Why the sensor sees what the operator cannot

A press set by hand is controlled from the outside: the operator sets a fill time, a pack pressure, a cool timer, and trusts that the plastic does the same thing every shot. It does not. Resin lots vary, the screw wears, the mold heats up over a run, and the process drifts. A cavity-pressure sensor sits inside the tool and measures what the plastic actually does, shot to shot. Once you know the pressure signature of a good part, you can auto-sort a bad one the instant it happens, at the press, instead of finding it in a box of finished parts hours later.

That is the point. Scrap caught at the press is one bad part. Scrap caught at the box is a whole run, plus the sort, plus the risk that some slipped through to the customer.

B. Scientific (decoupled) molding, in plain terms

"Decoupled" molding just means you stop treating fill, pack, and cool as one lumped action and control each on its own:

  • Fill the cavity fast and consistently (about 95% full), controlled by speed.
  • Pack it out to final dimensions, controlled by pressure.
  • Cool it only as long as the part actually needs, not a round number set with extra time to be safe.

You find the right settings with a DOE (a structured set of experiments), not by nudging dials. The result is a wide, robust window: the process can drift a little and still make a good part, instead of sitting one bad lot away from scrap.

C. The PPAP gate, and how to not wait months for the money

This is the catch that turns an eight-month payback into eighteen if you ignore it. A change to a validated process, tool, or material on a customer part generally requires notifying the customer and, unless they waive it, a re-approval (PPAP in automotive, IQ/OQ/PQ in medical) before the improved parts can ship, and the customer, not you, controls that gate. The way through:

  1. Segment the book. Validated parts vs. commercial. Prove the new methods on commercial parts first, where there is no re-validation cost at all.
  2. Monitoring-only on validated parts. Run the sensors to watch and sort without touching the approved set-points. That captures the scrap-catch value with no re-PPAP. Confirm the scope with the customer's quality engineer.
  3. Launch new tools scientifically from day one, so new work is born controlled and never needs a re-validation.
  4. Batch and piggyback. When you do re-validate, make all the improvements at once so you pay the PPAP toll once, and fold your changes into any engineering change the customer is already funding.

D. Prove it is worth doing before you buy a sensor

The entire case rests on the scrap being process-driven and the tools being sound. Pull the scrap tickets, sort process faults from tool faults, and rank the tools by scrap dollars and press hours. If the top faults are flash and drag, that is the tool-maintenance card's money, not this one.

What we did not have in the corpus and would request from the client on day one. The aggregated gaps across all 15 cards become the engagement-letter ask sheet.

  1. 01

    Scrap log by tool and defect code (last 3 to 6 months)

    Owner

    Quality manager

    To confirm the 5% scrap rate and, more important, to sort process faults (fixable here) from tool faults (fixable by the mold-maintenance card). This one dataset decides whether the card is worth building and which tools to instrument first.

  2. 02

    Resin purchased vs. good parts shipped, by weight

    Owner

    Materials / purchasing

    To ground the true, by-weight scrap rate rather than the reject count molders usually quote, since every dollar on this card scales off it.

  3. 03

    Customer PPAP files and control plans for the pilot parts

    Owner

    Quality manager

    To scope the re-validation on each pilot part and decide which stay on monitoring-only versus which are worth a full delta PPAP.

  4. 04

    List of presses and tools with cavity-pressure capability

    Owner

    Tool room / process engineering

    To know which tools can take sensors as-is and which need work first, so the $80K lands on tools that can actually hold a signature.