Value Stream Mapping for Float Glass Manufacturing: From Batch Charge to Wrapped Pack Without the Ribbon Break Drag

Float glass is a different manufacturing environment from discrete production. The ribbon does not stop when a downstream order changes. The furnace runs continuously, typically for 12–15 years between rebuilds, and the commercial result depends less on making the line faster than keeping it flat, straight, stable and predictable.

That is why Value Stream Mapping (VSM) is so valuable. It makes waiting, inventory, defects, information delays and recovery losses visible across a process that appears to be one uninterrupted flow.

This worked example maps a product family from batch charge to wrapped pack and shows how a future state can improve lead time, yield, OEE and delivery reliability without disturbing the physics of the glass process.

1. Define the Value Stream Boundary

The scope begins at raw material receiving, batch weighing and mixing, then follows the material through:

1. Furnace charge and melting
2. Float bath and molten tin forming
3. Annealing lehr
4. Online inspection and defect marking
5. Cold-end cutting and snapping
6. Edge working, including arrissing and seaming
7. Washing
8. Tempering or laminating where required
9. Packing, racking and warehouse staging
10. Dispatch to the customer

The product family is:

• 4–6 mm clear float glass
• 4 mm low-iron float glass

These products share the principal process route and account for approximately 78% of the line’s volume, making them suitable for one family-level VSM.

Takt time versus line rate

Average customer demand is 372 tonnes per day, or:

[
372 \div 24 = 15.5\text{ tonnes per hour}
]

At a 3,600 mm-wide, 4 mm ribbon, the theoretical mass is approximately:

[
3.6m \times 0.004m \times 2,500kg/m^3 = 36kg\text{ per metre}
]

At an 18-tonne-per-hour pull rate:

[
18,000kg \div 36kg = 500m\text{ per hour}
]

Customer demand requires approximately:

[
15,500kg \div 36kg = 431m\text{ per hour}
]

The line therefore runs at 500 metres per hour while the customer demand rhythm is about 431 metres per hour. The VSM question is not simply, “How can we increase speed?” It is, “How can we synchronise production, cutting, packing and dispatch with demand while protecting furnace and ribbon stability?”

Customer-defined value consists of:

• Consistent thickness and flatness
• Optical quality with minimal distortion
• Cut-to-size accuracy
• Low defect and breakage rates
• On-time delivery
• Competitive price per square metre

The float process itself is well described by Pilkington’s float process guide and Guardian Glass’s manufacturing overview.

2. Current-State Map: Follow the Glass and the Information

Batch house and furnace charge

A representative 1,000 kg batch may contain approximately:

• 570 kg silica sand
• 150 kg soda ash
• 80 kg limestone
• 70 kg dolomite
• 130 kg cullet

Materials are received, stored, weighed and mixed before being charged continuously. The batch house holds approximately 1.2 days of raw material inventory. Weighing and mixing require around 3 minutes of active processing time per batch equivalent, but material may wait 4–8 hours for release, quality checks or furnace scheduling.

The furnace pulls 18 tonnes per hour. Furnace and float operations deliver an OEE of 71.4%, calculated as:

[
89%\text{ availability} \times 91%\text{ performance} \times 88.3%\text{ quality}
]

[
0.89 \times 0.91 \times 0.883 = 71.5%\approx71.4%
]

The furnace consumes approximately 7.9 GJ per tonne of glass, with the furnace absorbing most of that energy. A stable batch recipe, consistent cullet ratio and controlled combustion profile are therefore central to both quality and energy performance.

Float bath, tin bath and annealing lehr

Molten glass forms a continuous ribbon on the tin bath. The ribbon is approximately 3,600 mm wide and is pulled at 18 tonnes per hour. Forming and controlled cooling through the lehr represent approximately 12 minutes of value-creating process residence time in this model.

The furnace, bath and lehr operate continuously, but a grade change from clear float to low-iron float currently creates 42 minutes of off-specification production. The line records three ribbon breaks per month, each requiring an average of 5.5 hours to recover.

That equals:

[
3 \times 5.5 = 16.5\text{ hours of recovery time per month}
]

The physical ribbon may be continuous, but the value stream is interrupted by recovery, stabilisation and downstream replenishment delays.

Inspection and defect marking

Laser scanners and camera systems identify bubbles, stones, tin pickup, ream, distortion and surface marks. Defects are marked for removal or classification at the cold end.

Current quality loss is 1.1% of input glass through off-quality material rejected at inspection. Inspection data are recorded, but defect codes do not consistently reach furnace or bath operators within the same production shift. This delay weakens the connection between the Voice of the Process and corrective action.

Cutting, snapping and edge working

The cold end cuts the ribbon into customer or stock dimensions, snaps sheets and removes trim. Arrissing or seaming is completed where the specification requires worked edges.

Total physical yield loss is 4.7%:

• 2.9% trim and edge loss
• 1.1% off-quality rejection
• 0.7% handling breakage

The arithmetic is:

[
2.9% + 1.1% + 0.7% = 4.7%
]

At an 18-tonne-per-hour pull rate, the theoretical good output after physical yield loss is:

[
18 \times (1-0.047)=17.15\text{ tonnes per hour}
]

After accounting for campaign interruptions, recovery and changeover effects, effective sellable output is approximately 16.4 tonnes per hour.

Washing, tempering, laminating and packing

Sheets are washed, inspected again and sent to tempering or laminating where required. These downstream cells add approximately 7 minutes of active value-added processing, including cutting, edge work, washing and packing.

Packers build pallets, add interlayers, apply corner protection and label finished goods. Finished stock averages 6.5 days of production. That inventory is not merely a storage figure: it represents customer demand produced before the call-off is available.

The current lead-time ladder is:

• Raw material and batch waiting: 0.4 days
• Furnace-to-cold-end flow and quality release: 0.3 days
• Finished goods warehouse: 6.5 days
• Dispatch and customer call-off: 0.9 days

Total lead time:

[
0.4+0.3+6.5+0.9=8.1\text{ days}
]

Total true value-added time is approximately 22 minutes:

[
\text{PCE}=\frac{22}{8.1 \times 24 \times 60}\times100
]

[
\text{PCE}=0.188%\approx\text{under }0.2%
]

3. The Eight DOWNTIME Wastes in Float Glass

4. Future-State Design: Stabilise Before Accelerating

The future state protects the continuous assets while improving the cold-end flow.

Furnace and ribbon stability

Create standard charge conditions covering:

• Batch recipes and moisture limits
• Cullet ratio control
• Furnace pull-rate windows
• Stove and combustion settings
• Tin bath atmosphere checks
• Lehr temperature-profile verification

A defect feedback loop should route inspection codes to furnace and bath operators within five minutes, rather than waiting for a later review.

Cold-end TPM

Use Total Productive Maintenance (TPM) for cutters, washers, conveyors and labelling systems. Operator checklists should verify blade condition, conveyor alignment, washer pressure, sensor cleanliness and emergency-stop readiness at the start of every shift.

The objective is not more inspection after a stoppage. It is earlier detection of deterioration.

Quick grade changes and trim optimisation

Pre-stage settings, standardise the thickness-change sequence and separate internal preparation from external machine time. The target is to reduce grade changes from 42 minutes to under 20 minutes.

For cut-to-order production, use nesting rules that match customer dimensions directly to ribbon width. This reduces avoidable trim without changing the furnace pull.

Packing and dispatch standard work

Use pallet-build photographs, interlayer specifications and corner-protection standards. Then level customer call-offs by dock hour, using a load plan that matches finished-pallet availability with transport capacity.

This creates a more reliable pull signal from dispatch back to cutting and packing.

The current-state map, kaizen actions and metric table can be maintained digitally in Ci Flow, which provides value stream mapping, project governance, tool evidence, actions and portfolio reporting in one operational-excellence workspace.

5. Current State versus Future State

Future-state output is based on:

[
18 \times (1-0.026)=17.53\text{ tonnes per hour}
]

With improved campaign stability and fewer recovery losses, this rounds to approximately 17.6 tonnes per hour of effective sellable glass.

6. 30/60/90-Day Kaizen Sequence

Days 0–30: Establish control

Owners: Furnace Operations Manager, Quality Engineer, Cold End Team Leader

• Confirm the product family, scope and measurement definitions.
• Validate the 4.7% loss breakdown by shift, grade and defect code.
• Introduce five-minute inspection-to-process defect escalation.
• Publish standard batch, cullet and combustion conditions.
• Begin daily ribbon-break Pareto reviews.
• Photograph and approve the standard pallet build.

Milestone: verified baseline and visible daily management board.

Days 31–60: Improve flow

Owners: Maintenance Planner, Cold End Team Leader, Logistics Supervisor

• Launch autonomous maintenance checklists for cutters, washers and conveyors.
• Pre-stage grade-change tools and settings.
• Pilot a sub-20-minute thickness-change sequence.
• Test cut-to-order nesting on the highest-volume dimensions.
• Reduce duplicate rack-to-dispatch movements.
• Introduce dock-hour load levelling.

Milestone: first future-state pilot with measured yield, changeover and dispatch results.

Days 61–90: Sustain and govern

Owners: Continuous Improvement Coach, Furnace Operations Manager, Quality Engineer

• Lock standard work into shift audits.
• Review OEE, yield, lead time and on-time delivery weekly.
• Confirm the maximum one-break-per-month target.
• Remove obsolete finished stock and reset replenishment parameters.
• Run a 30-day control review using trend charts and defect stratification.
• Convert successful kaizen bursts into controlled projects.

Milestone: future-state map becomes the operating standard, not a one-time workshop document.

Build the Capability Behind the Map

A float glass VSM is most effective when the team can connect material flow to statistical evidence, TPM routines, OEE, defect data and control plans. That capability comes from structured Lean Six Sigma certification, not from drawing a map in isolation.

Lean 6 Sigma Hub offers CSSC-accredited training from White Belt through Master Black Belt. For professionals leading complex manufacturing improvement, explore the practical Lean Six Sigma Black Belt online training, including DMAIC, process mapping, statistical analysis, TPM-related improvement and advanced project leadership.

You can also review the full range of Lean Six Sigma online training and belt levels.

Choose the right belt level, map your value stream with evidence, and lead the improvement from current state to controlled future state.

Kaizen. Kai-Care. Kai-Done. Lean Six Sigma