The economics pillar is the cluster's hub. It models the per-can dollar figure that reclaimed residual has to clear before capital review will fund it. The field-proof pillar is the leg of the cluster that produces the inputs the model reconciles against — SKU-level residual data, line-side pilot yield, facility-specific throughput-band validation. This pillar is the third tier of that same cluster, and it is the tier neither of those two pieces owns.

Operations is the operating discipline the model and the field-proof package both depend on but neither covers. The model runs on trapped-product volume and on the yield that volume converts to recovered dollars. The field-proof package measures the volume and validates the yield. But the volume itself — how big it is, how it moves, what shifts it up or down month over month — is set on the fill line, by operating decisions, against operating data that the operations tier alone owns. Both the economics model and the field-proof package assume the operating tier has held. Neither of them enforces it. This pillar is the enforcement layer.

It is the third in the series rather than the first because the levers it covers were already named as a sibling tier in the economics pillar and the field-proof pillar before this piece existed. The operations tier does not introduce a new framework. It picks up the lever list that the cluster already works against and gives it a discipline of its own.

The Four Operating Levers — And Why the Cluster Has Only Ever Named These Four

The cluster's published lever list is closed. There are exactly four knobs that move trapped-product volume on an aerosol fill line, and every reclaim-economics case that gets funded rests on knowing where each of them is sitting.

Propellant ratio calibration is the first lever. Bulk propellant deliveries arrive at slightly different densities, ambient conditions shift through the day, and operators adjust ratios to compensate for downstream complaints they may not be tracking — collectively producing a 2–7 percentage point drift from the spec the line was commissioned against. The dollar figure the model produces is sensitive to where the ratio actually sits, not where the spec claims it sits.

Valve-to-product reconfiguration is the second. The valve on the can was chosen for a particular spray character against the propellant pressure the line ran when the SKU launched. Then the propellant supplier changes. The product formula adjusts for cost. The valve sits unchanged. A valve mistuned to the current fill profile leaks product both at fill (poor seating against the modified pressure curve) and at consumer use (spray geometry not matched to what's inside). Both contribute to trapped residual.

Overfill tolerance tightening is the third. Fill-weight specs are written with margin — 2 grams of nominal margin on a 250ml can plus whatever the checkweigher accepts. That margin is product, and a 1-gram tighter target on a 50-million-unit line recovers roughly 50,000 kilograms of product shipped as margin per year, most of which never reaches the consumer.

Changeover flush discipline is the fourth. Lines running frequent changeovers for production scheduling reasons accumulate flush loss that looks like a constant cost but is actually proportional to schedule complexity, not throughput. Capture-and-return systems and pre-staged batch transitions reduce the volume lost in each changeover without new capital equipment.

Each of these four levers produces its own operating-data artefact — and the four artefacts are the inputs the field-proof pillar assigns to this tier. The cluster has never named a fifth lever because no fifth lever has ever produced meaningful trapped-product movement on a line that has the other four in calibration. Adding a tier-five lever to a model that has ignored the existing four is a vendor pattern. Operating discipline is built on this list of four, in this order.

The Operating-Data Artefacts Each Lever Produces

Each operating lever has a specific data artefact — the kind of running record that proves where the lever is sitting and what changed when it was tuned. Without these artefacts, the lever has been moved by feel. With them, it has been moved by measurement.

Propellant calibration log. Recorded at the start of every shift, against the spec, against the can weight post-fill. Most lines that start logging find drift they did not know existed — typically 4–7 percentage points from spec on any given production run. The log becomes the dataset the lever is adjusted against and the artefact the economics model reconciles between quarterly reviews.

Valve configuration audit. A per-SKU registry of which valve is currently mounted against which fill profile, dated, with the propellant ratio, product formula, and any downstream change events captured. The audit reveals where the valve no longer matches the can — and where replacement or reconfiguration closes trapped-product loss at the moment the line would otherwise have run with a mismatch.

Fill-weight distribution. Quarterly pull from the line's checkweigher output, comparing actual fill-weight distribution against the spec band and against the published tolerance stack. The distribution reveals where overfill margin is sitting — and what a 1-gram tighter target produces against the acceptance rate. The economics pillar uses this distribution as one of the inputs that constrains the per-can recovery band.

Changeover flush log. Flush loss as a percentage of total product throughput, tracked weekly. Most operations that start tracking find a number substantially higher than expected — and that is the lever. The log is what converts flush loss from a price-of-doing-business number into a controllable operating input.

These four artefacts together compose the operating-data layer that the field-proof pillar's package is tuned against. Without them, field-proof has measured residual but no validation against what is driving it. With them, the field-proof package can defend the cause, not just the effect.

How Operating Data Feeds the Field-Proof Package

The field-proof pillar assigns four artefacts to the operations tier. This pillar owns the production of those four artefacts — and the economics pillar downstream relies on them too.

The taxonomy that links the two tiers is direct. Propellant calibration log → input to yield validation at facility-specific propellant ratio. Valve configuration audit → input to the SKU-level residual measurement's segmentation by valve geometry. Fill-weight distribution → input to the throughput-band validation's stability check at the operating tolerance the line actually runs at. Changeover flush log → input to the throughput-band validation across the SKU schedule the line is expected to run in the capex year.

What this means in practice: a field-proof package that arrives at capex review without the four operating artefacts is a package that has to be argued from the residual end alone. The committee can fund the residual reduction, but it cannot fund the operational discipline that maintains it. The commit will be made against a residual number whose cause has not been documented, and the sustainment of the gain becomes a fight the operations team has to have separately.

A field-proof package that arrives with the four artefacts is a different case. Cause and effect are documented line by line. The committee can fund the residual reduction, the operating discipline that produced it, and the same operating discipline that is expected to sustain it. The fight is one fight, not two.

Sequencing Tuning Without Disrupting the Rest of the Line

The four levers compound with each other. Tuning two at once produces a delta that cannot be attributed, and an attribution gap is a year-one problem on the capex review. The methodology for handling this is sequential, not parallel.

Establish a baseline first. Pull three months of fill-weight distributions, propellant ratio logs, changeover counts, and any end-of-line residual data the line already has. Most of this exists somewhere on the line — it just has not been assembled into a single view. The baseline is not a number; it is a panel of four running records, integrated into the same time window.

Then run the obvious diagnoses. Where is the propellant ratio actually sitting against spec? Where does the overfill margin exceed what is needed? Which SKUs have the highest residual signature at the end-of-line checkweigher? The diagnoses do not move any levers. They identify where the lever movement is likely to land and where the dollar figure is most likely to grow.

Then tune one lever at a time. Adjust propellant calibration on one line, hold everything else constant, measure the delta against the four operating-data artefacts. Each lever compounds with the others, but isolating them for the first cycle is how attribution is preserved. Two levers moved over the same quarter produce a delta the committee will not be able to defend disaggregated.

Then track dollar recovery, not just percentage. A 1-percentage-point reduction in fill loss on a 50-million-unit line is a real number, in real dollars, that finance will recognize — and the dollar figure is the unit the per-can recovery band reconciles against. Speak in those terms when the operating discipline's gain is being advocated for.

The sequencing matters because the capex committee is reviewing the case at a single point in time. Operating-discipline work that is sequential and documented produces a defensible case at that moment. Operating-discipline work that is parallel and undocumented produces a case the committee has to take on partial trust.

Measurement Discipline for Sustaining the Gain

A one-time tuning is not a programme. It is a calibration run.

The four operating-data artefacts are running records. If the propellant calibration log stops at the end of the tuning cycle, the next propellant supplier change will drift the ratio back without anyone noticing. If the valve configuration audit stops, the next product-formula adjustment will create mismatches that the line will leak through until the next audit. If the fill-weight distribution stops being pulled quarterly, the overfill margin will drift back to the historical baseline within a quarter or two as acceptance rate patterns stabilize on historical tolerance. If the changeover flush log stops being weekly, the schedule complexity will find its way back into flush loss without anyone naming it.

The sustaining discipline is the same panel of four running records, kept current indefinitely, rather than assembled once and shelved. The economic stake on keeping them current is the gap between the residual number the model assumes the line is running at and the residual number the line is actually running at.

A facility that has tuned once and stopped tracking will find within two quarters that the trapped-product volume has drifted back to within a point or two of the pre-tuning baseline. The model the case was funded on is invalidated. A second tuning cycle is required to restore the case — and the second cycle is more expensive than the first because it has to be argued as a recapture, not a fresh initiative.

A facility that has tuned once and kept the four running records current will find the trapped-product volume at the post-tuning level, quarter over quarter, with drift identified and corrected within weeks rather than quarters. The model's inputs remain aligned with the operating reality. The capex case sustains itself.

The operating-data layer is what converts a published cluster number — "12–15% residual, 60–85% recoverable" — into a facility-specific number that holds across the capex year. Without the layer, the cluster number is an anchor that drifts; with it, it is a number that tracks.

Reconciling Operating-Layer Gains Into the Per-Can Number

The economics pillar walks the per-can cost recovery band on a 50-million-unit line: $0.18 to $0.42 per can, summing yield recovery ($7.2M–$12.75M), disposal-fee avoidance ($3.5M–$4.5M), and EPR eco-modulation ($500K–$1M). The operating-layer lever that lands those numbers is the 1–3 percentage point movement in trapped-product volume — the band the four levers typically produce in a tuned-versus-untuned comparison on a facility that has been running the operating programme for a quarter or a year.

One percentage point of trapped-product reduction, applied to a 50-million-unit line at $2.00 average per-unit value and an 70% midpoint yield, is roughly $1.0M annually on the yield-to-dollar layer alone, before disposal-fee avoidance or eco-modulation. Three percentage points is roughly $3.0M annually, with the eco-modulation surcharge layer adding another $300K–$600K on a high-residual profile the year measurement begins. The operating tier is what makes that 1–3 percentage point movement real rather than asserted.

A reclaim case that arrives at capex review without the operating-data artefacts will be challenged on the per-can figure's defensibility. The committee will ask: if the per-can band is $0.18–$0.42, where on that band is this facility — and what has been measured to defend the upper-quartile assumption? The operating tier's four artefacts are exactly the defence that question requires. A 1.8-percentage-point measured trapped-product reduction, documented across three quarters of the four running records, defended panel by panel — that is what produces a per-can figure that survives capex review intact.

What an Operations-Defensible Package Looks Like Before the Field-Proof Programme Starts

The field-proof pillar ends with the package a capex committee funds. This pillar ends on the package the field-proof programme needs to find when it arrives line-side — the operations-defensible package that has to be in place before SKU-level residual measurement begins.

Three readers, three different questions:

- Operations. "Is the lever set in calibration, or does measurement arrive at a line running on historical tolerance?" The package has to show all four operating-data artefacts running current at the time measurement begins. Otherwise the measurement baseline includes drift the operating tier would have already addressed. - Finance. "Is the per-can figure derived from running operating data, or is it derived from spec?" The operations-defensible package has to show the per-can inputs that the economics pillar walks through — trapped-product volume, propellant ratio actuals, valve configuration status, overfill margin actuals, changeover flush percentages — all drawn from the four running records. - Capex committee. "Before the field-proof programme adds its own measurement cycle, is the operating baseline honest?" The operations-defensible package assumes the answer is yes, with every panel of the four running records running current and reconciled against the same time window the field-proof tier will measure against.

What goes in the package:

- A propellant calibration log running at least ninety days, with at least three shifts recorded per day, against the spec the line was commissioned against. - A valve configuration audit covering every SKU the line runs in the prospective capex year, dated, with any product-formula or propellant-supplier change events captured. - A fill-weight distribution pulled from the last quarter's checkweigher output, with the published tolerance stack reconciled. - A changeover flush log tracking flush loss as a percentage of total product throughput, weekly, across at least one quarter of operation.

What does not go in:

- A snapshot of lever position from one day, one shift, or one SKU. - A valve configuration audit that has not been updated since the last product-formula change across any SKU. - A fill-weight distribution pulled longer than six months ago, before the current tolerance stack. - A changeover flush log covering fewer than twelve weeks, before the schedule complexity pattern visible across a full quarter has emerged.

The operations-defensible package is what the field-proof programme builds on. Without it, field-proof is calibrating against an operating line that may still be drifting. With it, the field-proof measurement catches the operating layer already in calibration — and the gap it measures is the gap between in-calibration operation and full reclaim yield, not the gap between drift and operation.

Pulling the Operations Tier Back to the Cluster

The cluster is three pillars and a live calculator. The economics pillar Aerosol Reclaim Economics: The Pillar Guide models the per-can figure. The field-proof pillar Aerosol Measurement and Validation builds the capex-defensible package. This pillar — the operations tier — owns the operating-data discipline without which the model's inputs cannot be defended and the field-proof package cannot be sustained across the capex year. The cluster's calculator at `/resources/manufacturer-roi` runs the same five-layer model the pillars describe, on the inputs an operations-defensible package has produced.

The order the cluster reads is operations first, economics second, field-proof third. The economics pillar coordinates the model. The field-proof pillar builds the capex case on top of it. The operations tier is what makes both function. Pick the entry point that matches the question your team is asking — but the cluster only holds together if all three pillars are in place.

Run the per-can math on your facility →

Subscribe for more guides →

FullCan helps aerosol manufacturers measure, reduce, and recover product waste. Built for operations teams who need numbers, not narratives.