Container Glass Total System Cost: 24 APAC Projects, What Stuck

24 Total System Cost projects across Australia, Vietnam, Indonesia and Thailand over about a decade, and the recurring savings landed in four buckets. Not 50 of them. Four. Finance teams love a 50-line tracker with a champion assigned to every row. Real plants don't run that way, and TSC delivery that pretends otherwise burns out the people who have to sustain it after the consultant's gone home.

What Total System Cost actually measures on the hot end

Total System Cost isn't a furnace cost line. It's the cost per saleable tonne once you count everything downstream of the melt: forming yield, packaging efficiency, cullet return, secondary rejects showing up on a customer's filling line. A plant can hit its melting cost budget every month and still bleed TSC through a forming section nobody's watching closely enough.

On the hot end that means gob weight CV held at 0.4% or better, forehearth profile within ±2°C across five zones, and a mould shop that can hand you a preheat curve at 480°C ±10°C without an argument. Miss any one of those and the TSC number moves, even when the melting cost report looks clean.

Where the 24 projects actually found the money

Split the 24 projects by where the recurring dollars actually showed up 12 months later, not at the workshop readout, and you get four buckets.

• Cullet quality and ratio
• Yield and downstream breakage
• Energy per tonne melted
• Job change scrap and first-ware loss

Cullet was the bigger lever more often than people expect. Lifting cullet ratio from the mid-40s to 55-60% on a container glass line, when the supply chain can actually support it, typically pulled energy intensity down 3-6% and cut blister rejects that everyone had been blaming on furnace atmosphere. Check the cullet contamination log before you touch the burner map (and yes, I know the furnace engineer swears it's atmosphere, check it anyway). That's usually where the stones and the blisters actually start.

The job change bucket surprised the plants that had never systemised it. Cross-shift variance on identical SKUs runs 30-60% in plants still running changeovers off a notebook, and that variance shows up in TSC as first-ware scrap and stabilise-stage downtime. This is exactly the gap our systemised Job Change Tool is built to close, and it's vendor-neutral by design, because half the plants I've worked have a mixed-OEM fleet and no appetite for another OEM-tied programme. Where the discipline held, sustained job change time fell close to -60% against baseline, and it recurred every quarter, not just in the pilot line.

The recurring savings live in cross-shift discipline, not the kickoff workshop

In 2018 I ran a TSC engagement on a two-furnace, five-line plant outside Jakarta. Old fleet, mostly 1990s-generation Emhart machines with pneumatic timing, no electronic NIS retrofit. The kickoff workshop identified $4.2M AUD in opportunity across the site. Eighteen months later, the sustained number was closer to $1.6M AUD. The gap wasn't bad analysis. It was ownership.

The 0600 handover sheet on that plant logged swabbing frequency maybe three days out of ten. Nobody owned the number once the project team left. The hot-end superintendent owned recipe lock on paper, but the operator was still adjusting plunger set points on night shift without sign-off, chasing count instead of holding spec.

And that's the pattern across most of the 24. The projects that held their number had a named owner for every recurring line item, reviewed weekly, with the KPI visible on the floor, not buried in a monthly deck. The ones that decayed had a project closeout report and nothing else.

Total System Cost only means something once someone's name is on the recurring number, not the one-time win in the readout deck.

What sustains a TSC number after the consultant leaves

A management audit six months post-project catches the drift before it compounds. We now run this as a standalone service because it kept coming up as the missing step: someone independent checking whether the meeting cadence, the KPI ownership and the shift handover discipline are still doing what the TSC business case assumed they'd do. Our management audit exists because of exactly this failure mode, watched across enough sites to build a service around it.

A GCC operator I audited last year had the opposite problem to the Jakarta plant. New capacity, strong Vision 2030-linked investment, but a mixed-OEM fleet built out fast with heavy turnkey dependence on the original equipment vendors. Every TSC recommendation from the commissioning contractor pointed straight back at that vendor's own upgrade catalogue. The plant management team wanted numbers they could defend at board level from someone with no product to sell them, which is most of what a vendor-neutral container glass consultant is actually there to provide.

Look, the spreadsheet always says the project worked. The floor tells you whether it stuck.

Zaid built the TSC discipline at O-I from the inside before he ran a single client engagement, and his TSC Champion experience is the reason Lean Glass treats sustainment as a separate deliverable instead of an afterthought slide.

So pull your last TSC readout and check one thing: 12 months on, does anyone still own the recurring number, or did it quietly become last year's project?