The holding time was right. The part still needed material.

On one of our ear­lier projects, a cover molded in a two-cav­ity tool had sink marks on its vis­i­ble sur­face. The hold­ing time had been set by a gate-freeze study: part weight stopped ris­ing once hold­ing pres­sure was held for longer than two sec­onds, and that value had gone into the process sheet.

The mea­sure­ment was cor­rect. The ques­tion was why pack­ing stopped while the part still needed mate­rial.

The orig­i­nal sim­u­la­tion had pre­dicted a longer effec­tive pack­ing time, but it mod­eled a dif­fer­ent geom­e­try at the gate. In the tool, the gate sat on a dome, an aux­il­iary geom­e­try on the core side, so the wall directly at the gate was at its thinnest. This sec­tion froze early and closed the pack­ing path. Once it had frozen, a longer hold­ing time could not push more mate­rial into the part.

We milled the dome off, repeated the gate-freeze study and set the hold­ing time to suit the mod­i­fied geom­e­try. The longer hold­ing time now reached the part, and the sink marks on the sur­face dis­ap­peared.

One mark remained: a shallow dent directly at the gate.

It had been there from the start. Among sev­eral sink marks it looked like one of them, and the soap-water film used for inspec­tion smoothed it over; a stripe-light pro­jec­tor showed it clearly. Its loca­tion set it apart. The part is filled through a banana gate, a curved tun­nel gate that breaks off dur­ing ejec­tion. It had been built to draw­ing, but the draw­ing did not fol­low the design rules for this gate type, and the gate held the part with a high extrac­tion force. As it broke off, it pulled the under­side in, and the vis­i­ble side sank above it.

We replaced it with a stan­dard insert from a mold-com­po­nents sup­plier, started at 1.8 mm and opened it step by step dur­ing sam­pling to 2.4 mm, so that the seal time suited the part and the extrac­tion force stayed low. The dent was gone.

Two mechanisms in one workspace

I use this older case because it shows the work­ing logic behind SEM, the Strate­gic Engi­neer­ing Matrix. The method was devel­oped from projects like this one; the case is mapped to it ret­ro­spec­tively.

In a live inves­ti­ga­tion, sev­eral mech­a­nisms can carry the same symp­tom, and tool­ing, mate­rial and process work often run in par­al­lel. The matrix keeps the mech­a­nisms and the find­ings assigned to them in one workspace: what sup­ports a mech­a­nism, what speaks against it and what still has to be checked. Each new find­ing changes that stand­ing. In the cover case, resolv­ing the pack­ing path did not close the inves­ti­ga­tion; it made the sec­ond mech­a­nism vis­i­ble. Work­ing on sev­eral mech­a­nisms in par­al­lel is dif­fer­ent from chang­ing sev­eral process set­tings at once with­out a plan. Every trial still needs recorded con­di­tions and a ques­tion it can answer.

Choos­ing an inter­ven­tion depends on the sit­u­a­tion at the plant. A tool cor­rec­tion may be tech­ni­cally right and still take longer than the next deliv­ery allows.

SEM dis­tin­guishes three types. An interim inter­ven­tion secures deliv­ery while the mech­a­nism is still open; it needs clear lim­its and mon­i­tor­ing. A robust­ness inter­ven­tion widens the process win­dow once the mech­a­nism is known, within an oper­at­ing range that has actu­ally been checked. An opti­mal solu­tion elim­i­nates the mech­a­nism. These are not stages to work through in order. Each option is assessed by its reversibil­ity, fea­si­bil­ity, risk and effect on the mech­a­nism, and the con­text decides: time to the next deliv­ery, bud­get and who can approve the change. In the cover case, both cor­rec­tions elim­i­nated their mech­a­nism, and the time to the next deliv­ery allowed it, so no interim inter­ven­tion was needed.

Each inter­ven­tion was checked against its own mech­a­nism. After the dome was removed, the repeated gate-freeze study had to show that pack­ing now con­tin­ued for longer. For the gate insert, the dent had to respond under com­pa­ra­ble process con­di­tions, with demold­ing observed directly. A part that looks bet­ter is an impor­tant obser­va­tion; the check has to show that the response matches the phys­i­cal expla­na­tion.

What project management and quality need

This is where the tech­ni­cal work becomes use­ful to project man­age­ment and qual­ity. They have to explain the sit­u­a­tion to the cus­tomer, jus­tify the next step and state what has been secured. The engi­neer­ing record gives them that directly: the observed defect, the mech­a­nisms and the find­ings that sup­port them, the selected inter­ven­tion, its ver­i­fi­ca­tion, and the open points and lim­its. In SEM Acad­emy we call this deci­sion doc­u­ment the Deci­sion Pack; SEM Desk­top gen­er­ates it as a final report or a one-page sum­mary. How it is used in the cus­tomer meet­ing is the sub­ject of a later part of this series.

SEM grew out of expe­ri­ence across more than 200 com­plex indus­trial projects. SEM Desk­top holds the matrix and links find­ings, inter­ven­tions and ver­i­fi­ca­tion; in SEM Acad­emy, we teach engi­neers to do that work. Another engi­neer should be able to pick up the case, fol­low the deci­sions and con­tinue from the same record.