Guide to the Processing Workflow of Custom Lunch Boxes | Design, Prototyping, Production

First, select food-grade PP/304 stainless steel materials with a capacity of 500-1000ml, and use CAD to design a 3D model with optimized parting lines;

Then, manufacture the mold using electrical discharge machining/3D printing, and perform trial injection molding (200-230℃, 80-120MPa) to verify the sealing;

Finally, CNC machine the container body, laser engrave the logo, conduct airtightness testing with a leakage rate of <0.5%, and use EPE shockproof packaging.

Design

Statista 2024: 85% of American consumers value rational compartmentalization, 70% focus on food-grade materials (PP/Tritan), 65% care about portability (handle width 4-5cm optimal).

Nielsen 2023: Layered design improves storage efficiency by 40%, magnetic clasp opening/closing rated 4.7/5.

Visual Design

Don’t be greedy with colors, 2 main colors + 1 accent color is most stable

Users process color in less than 0.3 seconds. With over 3 main colors, recognition rate drops 37%. How do international brands choose?

• Main Color: Must be tied to the brand’s VI color. For example, American outdoor brand Patagonia uses its iconic “earth yellow,” users say “I know it’s my bag from far away.” Statista tests show main visuals using brand colors have a 41% higher memory retention rate than random colors.
• Accent Color: Choose low saturation colors (e.g., off-white, light gray, pale blue), not exceeding 1/3 of the main color area.
• Pitfalls to Avoid: Avoid high-saturation clashing colors (e.g., bright red + bright yellow), 45% user feedback says “looks like a child’s toy”; dark backgrounds (black, dark blue) with light patterns show dirt easily, REI tests show stain visibility is 2 times higher than light backgrounds.

LOGO placement and size, decided by user’s eyes

The natural focal point for user’s gaze is the top right corner of the front, placing it there increases click-through rate by 29% compared to the center.

• Size: LOGO height about 1/5 of the box face is optimal. For example, American brand Yeti’s custom models, LOGO height 4cm (box face length 20cm), users say “not too small and doesn’t block view”; smaller than 2cm (<1/10 of face), 32% user feedback says “can’t find it”.
• Contrast: Color contrast between LOGO and background must be sufficient. White background with black LOGO recognition rate 98%, similar color schemes (e.g., light blue background with dark blue LOGO) drop 21% (Nielsen 2023).
• Distortion: American retailer Target once compressed its circular LOGO, leading to user complaints “looks like a counterfeit,” return rate increased 18%.

Don’t make patterns with complex gradients, they’ll print blurry

Complex gradients (like rainbow transitions) have a 63% printing error rate (American Printing Association 2024 report), better to use simple graphics.

• Recommended Patterns: Geometric lines (straight lines, triangles), brand IP sketches (e.g., American coffee brand Blue Bottle’s bird IP), repeating small icons (e.g., leaves, stars). REI uses a “mountain line” pattern, user rating 4.7/5 (“has layers from afar, not messy up close”).
• Text: Brand slogan font size ≥8pt (use 10pt for smaller box faces), choose sans-serif fonts (e.g., Arial)—serif fonts (e.g., Times New Roman) have stroke adhesion at small sizes, recognition rate 19% lower.
• Case Study: American brand S’well transparent model uses a “single wavy line” wrapping around the body, cost 40% lower than full printing, users say “simple but distinctive”.

Choose the right process, texture differs by a factor of two

Three commonly used processes internationally, let the data speak:

For example, German brand ZWILLING uses hot stamping for its LOGO, users say “feels premium to the touch, not plasticky”; while screen printed versions are cheaper, 25% user feedback says “paint peels after long use”.

Different scenarios call for different visual styles

Office, outdoor, student models – what’s the visual difference?

• Office: Needs to “blend in discreetly”. US Amazon Basics models use “light gray + white”, users say “doesn’t stand out on the desk, colleagues ask for the link”; avoid bright colors (e.g., neon green), 38% user feedback says “looks like a kid’s lunch box”.
• Outdoor: Stain-resistant + visible. REI hiking models use “army green + earth yellow”, surface water-repellent treatment (mud wipes clean), user rating 4.7/5.
• Student: Fun but not childish. Models sold in California school campus stores use “sketch animals + low saturation colors” (e.g., light brown bear + off-white), parents say “kids willing to carry it, not think it’s ugly”.

Structural Design

Choose the wrong opening/closing method, and it’s annoying to use

International brand tests on 5 opening/closing methods show clear data differences:

For example, magnetic clasp models, users say “can open with one hand on a crowded subway, soup doesn’t spill”; but strong magnet versions occasionally attract keys, later brands added magnetic shielding (cost +$0.1), complaints dropped 40%.

Messy compartments like a warehouse, who would want to carry it

Statista 2024 survey: 83% users abandon repurchase due to poor compartmentalization. Layers + independent compartments are the foundation. How to divide specifically?

• 3-layer division is most versatile: Top layer for staple food (rice/pasta), middle layer for dishes (separating plate prevents flavor mixing), bottom layer for fruits/snacks. US Amazon Basics models use this design, storage efficiency 55% higher than single-layer (packs the same items 2 minutes faster).
• Don’t omit the independent sauce compartment: 5cm×5cm small compartment (2cm deep), perfect for salad dressing/chili oil. Target tests show adding a sauce compartment led users to say “don’t need to carry a separate small bottle, saves space”, rating increased 0.8 points.
• Soup compartment must be leak-proof: 8cm diameter circular deep compartment (4cm deep), with silicone seal. REI outdoor models use this, soup doesn’t leak after 1 hour inverted, user rating 4.9/5.
• Utensil compartment is a detail hider: Foldable spoon/fork (15cm long) + napkin slot (10cm×8cm), compartment door uses magnetic closure (quieter than clasp).

Handle and weight determine the willingness to carry it out

Nielsen 2023 data: Weight over 500g increases user abandonment probability by 62%; handle width <4cm, complaints of hand pinching 38%.

• Weight control: PP material wall thickness 1.0-1.2mm (too thin cracks easily), total weight 400-450g optimal. For example, IKEA models weigh 420g, users say “feels like carrying nothing in the bag”.
• Handle design: Curved shape (fits palm), width 4.5-5cm (distributes pressure), material uses soft silicone (anti-slip + doesn’t pinch). REI tests show 5cm wide handle reduces hand pressure by 57% compared to 3cm, rating 1.1 points higher.
• Add non-slip feet at bottom: Silicone pads (2mm thick), doesn’t slide on table. US S’well models added this, users say “doesn’t tip over during sudden car braking”, positive review rate increased 23%.

How structure differs for different scenarios?

• Office: Opening uses magnetic clasp (quieter than snap), inner liner uses PP removable (Target model inner liner washes clean in 10 seconds after removal), side adds pen holder (for temporary handwritten notes). User rating 4.8/5 (“can eat secretly during meetings without noise”).
• Outdoor: Corners add bumpers (ABS engineering plastic, impact-resistant), body has a hanging loop (1cm diameter, ties to backpack side pocket). REI hiking models use this design, drop test (1 meter onto concrete) 10 times no cracks, users say “just throw it in the bag when hiking”.
• Student: Clasp adds double silicone seal (doesn’t spill when inverted), rounded corners (R3mm radius, prevents knee bumps). Models sold in California school campus stores using this design led to a 70% drop in parent complaints about “soup spills”.

Choosing the Right Material

First look at safety certifications, don’t gamble with health

Internationally, food contact materials are strictly regulated, FDA (USA) and EC 1935/2004 (EU) are the baseline.

• FDA 21 CFR 177.1520: Specifies PP (polypropylene) melting point ≥160℃, total migration ≤10mg/dm² (no harmful substance migration with hot food). For example, US brand S’well uses FDA-certified PP for inner liners, users say “no smell when holding 80℃ hot food”.
• EC 1935/2004: Requires material heavy metal (lead, cadmium) content <0.01mg/kg. EU brand Muji custom models are all tested to this standard, reports publicly available.
• Pitfall: Don’t trust verbal “food-grade” promises, look for certification numbers (e.g., FDA’s “F” starting code). A US brand once used uncertified recycled plastic, users complained “plastic taste in food”, return rate soared to 35% (Nielsen 2023).

PP Plastic:

• Temperature resistance: -20℃ (frozen yogurt) ~ 120℃ (microwave heating), 50% wider usable range than PS plastic (-10℃~80℃) (ASTM D638 test).
• Wall thickness: 1.0-1.2mm optimal. REI tests show 1.2mm wall thickness PP, dropped from 1 meter onto concrete 10 times no cracks (1.0mm version crack rate 22%), cost only $0.1/unit higher than 1.0mm.
• Cost: 0.5-1 per unit (including mold cost allocation), 60% cheaper than 304 stainless steel (Statista 2024).
• User feedback: Target high-frequency use PP models, users say “no scratches after half a year, doesn’t deform in dishwasher”, repurchase rate 28% higher than ordinary plastic.

304 Stainless Steel:

• Insulation: Double-wall vacuum structure (0.5mm wall + vacuum layer + 0.5mm), retains heat for 4-6 hours (soup drops from 90℃ to 50℃ at 25℃ room temp).
• Corrosion resistance: Resists acid/alkali (pH 2-12), holding tomato juice, lemon tea doesn’t rust. Compared to 201 stainless steel (higher manganese content), 304 stainless steel has 73% lower rust complaint rate (Nielsen 2023).
• Weight: 30% heavier than PP (500ml capacity ~550g), but user acceptance is high.
• Note: Inner wall must be polished (Ra ≤0.8μm), otherwise stains remain hard to clean. IKEA models after polishing, users say “cleans with sponge in 10 seconds”.

Tritan:

• BPA-free: Passes FDA, EU EFSA certifications, no BPA release at high temperatures (100℃). US brand Contigo uses Tritan for transparent models, user trust rating 4.9/5 (“feel safe packing food for kids”).
• Transparency: 92% (ordinary PC plastic 85%), 40% lighter than glass (500ml ~300g). S’well transparent model users say “as clear as glass, doesn’t shatter when dropped”.
• Impact resistance: 1 meter drop onto marble floor, breakage rate <5% (PC plastic 15%).
• Cost: 50% higher than PP (1.5-2/unit), but young users willing to pay for “transparency + safety” (Target data shows sales increased 33%).

Eco-friendly Materials:

• PLA: Corn starch based, industrial composting (above 58℃) degrades in 6 months. Cost 30%-50% higher than PP (0.8-1.5/unit), suitable for one-time events (e.g., conference gifts).
• rPET: Recycled plastic bottles remanufactured, reduces carbon footprint by 30% (compared to virgin PET). Transparency 80%, suitable for outer packaging (not direct food contact).
• Limitations: PLA sensitive to heat (resistance <60℃), cannot hold hot soup; rPET may have slight odor (requires 24-hour ventilation).

Prototyping

According to US Packaging Digest 2023 report, products that undergo prototype testing have a 52% lower failure rate in mass production compared to untested ones;

A California company used 3D printed prototypes to gather feedback from 50 parents; after adjusting the handle curvature, children’s comfort rating increased from 6.2 to 8.9 (on a 10-point scale).

Single prototype cost <80, but avoided 120,000 in mass production rework losses.

Prototyping Methods

3D Printing:

FDM uses heated plastic filament (e.g., PLA, ABS) layered deposition, single unit cost 50-300, accuracy ±0.2mm, produces samples in 24-48 hours, suitable for early shape modifications.

For example, New York startup SnackBox designed a curved box body, used Ultimaker S5 printer, layer thickness set to 0.1mm (too thick shows layer lines, too thin is slow).

After 3 iterations, settled on a curvature radius of 12cm (fits palm curve), total printing cost $210.

SLA uses UV light to cure resin, accuracy ±0.05mm, surface smooth like injection-molded parts, but cost higher (150-500/unit), suitable for confirming appearance details.

For example, California brand BentoBeauty used it to print cartoon relief box lids, resin choice DSM Somos WaterClear (transparency 92%), tested UV transmittance.

CNC Machining:

CNC uses computer-controlled milling machines to “cut” metal or plastic parts, accuracy ±0.05mm (human hair diameter ~0.07mm), suitable for verifying structural strength and functionality.

Cost 800-2500/unit, depends on material and complexity.

German brand ThermoLunch making vacuum-insulated lunch boxes, used CNC to mill 304 stainless steel box body (1.2mm thickness), compared insulation effect of 5mm vs 8mm vacuum layer: holding 60℃ hot water, after 4 hours 5mm layer dropped to 38℃, 8mm layer to 45℃, chose latter (cost +$0.4/unit).

Process flow: CAD modeling → generate G-code → clamp aluminum blank → mill shape (spindle speed 8000rpm, feed rate 500mm/min) → edge polishing to Ra 0.8μm (smooth touch, no burrs) → drill holes for hinge (tolerance ±0.03mm to prevent looseness).

Silicone Casting:

Silicone casting replicates multiple samples from a master mold (e.g., a 3D printed part), single unit cost 200-600, accuracy ±0.1mm, suitable for testing different accessories (e.g., sealing gaskets, handles).

California brand EcoBento making soft rubber sealing gasket, master mold SLA printed resin part, mixed liquid silicone (Shin-Etsu KE-1310ST) ratio (A:B=1:1), vacuum degassed, poured, cured at 60℃ for 2 hours.

Tested 5 hardness levels (Shore A 40° to 60°): 40° too soft leaks (tested inverted with 50ml water leaked 3 drops), 60° too hard to install (opening/closing force 4.2N), chose 50° (0 drops leakage, force 2.1N).

Casting mold lifespan ~50 cycles, after that silicone shrinks (size deviation >0.2mm), requires new master mold.

Manual Assembly:

No machines, just cardboard + glue + stickers to make prototypes, cost <$100, suitable for quickly validating spatial layout.

A UK design team making children’s lunch boxes, used 3mm corrugated cardboard to cut box body (22cm L × 16cm W × 8cm H), inner layer pasted aluminum foil to simulate reflective insulation, compartment dividers folded from cardstock (4.5cm deep), fit standard US meal tray (21×16cm) perfectly.

Test found original divider too shallow (3cm) couldn’t hold yogurt cup, problem solved after increasing height.

Vacuum Forming:

Vacuum forming uses heated plastic sheet (PET/PS) drawn onto a mold, single unit cost 300-1000, suitable for thin-wall shells (e.g., transparent lids).

US brand FitLunch making transparent window lid, sheet material 0.5mm PET (heat resistance 70℃), mold used 3D printed ABS male mold, heated to 160℃ softened, vacuum drawn, cooled, trimmed.

Tested UV blocking rate: untreated PET transmits 25%, after TiO₂ coating reduced to 5%, cost +$0.2/unit.

Three rules international teams follow for method selection

Specific practices for selecting tools by stage

• Just have sketches: Use 3D printing FDM for basic shape (<$100), confirm dimensions fit bag size (e.g., US school locker 30×25×15cm);
• Modify functional structure: Use CNC machining for key parts (e.g., hinge, clasp), test strength within $1500 (e.g., clasp insertion/extraction 1000 cycles no break);
• Trial multiple versions: Silicone casting replicate 5-10 samples (<$500), let users compare and choose best (e.g., handle curvature version A vs B);
• Finalize appearance appeal: Vacuum forming for transparent lid or colored shell (<$500), take product photos test market reaction.

Hidden details of material testing

• Heat resistance: Lid holding boiling water, PP deformation 0.3mm at 100℃ (acceptable), ABS deformation 1.2mm (not acceptable);
• Impact resistance: 1-meter drop test, Tritan box intact rate 98%, ordinary PP box 82% (ASTM D256 standard);
• Environmental requirements: EU EPAC standard requires recyclable material content >30%, single PP material scores higher than mixed plastics.

Standardized Process

Step One:

• Weight <280g (upper limit of average grip strength for ages 6-10, tests show >300g increases child’s willingness to put it down by 65%)
• Corner radius R ≥2mm (prevents bumps, measured with radius gauge, <2mm samples have 41% higher scratch probability in drop tests)
• Antibacterial rate >99% (against E. coli, using ISO 22196 test method, count after 24-hour incubation)
• Insulation duration ≥4 hours (holding 60℃ food, center temperature not below 40℃, measured with iButton DS1922L recorder, accuracy ±0.5℃)

Step Two:

• Lid closure gap 0.2-0.3mm (measured with caliper, too tight hard to open, too loose leaks air)
• Clasp insertion depth 3mm (too shallow falls off easily, too deep hard to pull out)
• Handle mounting hole distance from box edge 1.5cm (prevents deformation under load)

Step Three:

Make prototypes according to the model, different parts use different methods, calculate cost and time separately:

German brand ThermoLunch making vacuum-insulated box, box body CNC milled from PP plate (1.5mm thick), compared 5mm vs 8mm vacuum insulation layer: holding 60℃ hot water, after 4 hours 5mm layer dropped to 38℃, 8mm layer to 45℃, chose latter (cost +$0.4/unit).

Step Four:

Three categories of testing, use tools to quantify results:

• Functional Testing
  • Insulation: iButton recorder measures center temperature every 30 minutes (target 4h ≥40℃), simulated Chicago winter outdoor (0℃ environment), single-layer PP box 32℃ after 4h, vacuum layer box 51℃
  • Sealing: Inverted holding 50ml water, shake 10 times (ISTA 3A standard), 0 drops leakage passes (a brand prototype leaked 3 drops, fixed after adjusting seal hardness)
  • Opening/Closing Force: Mark-10 M3-50 force gauge measures clasp (target <3N, equivalent to light mouse click), tested by 10-14 year old students one-handed, failure rate >30% requires spring-assisted structure modification
• User Field Testing
  • Sample: 20 US families (East/West coast, covering different climates), use for 1 week
  • Record: Video usage scenarios (one-handed opening, shaking in backpack), count “sticking incidents”; SurveyMonkey collects subjective scores (1-5 points, “handle comfort” improved from 3.1→4.6)
  • Finding: 73% children rotated lid to find clasp (took 8 seconds), added tactile bumps reduced to 2 seconds
• Durability Testing
  • Drop: 1.2m height free fall (ASTM D256), round-corner box (R3mm) breakage rate 9%, square-corner box 38%
  • Friction: Lightly rub surface with sandpaper (400 grit) 50 times, antibacterial coating peeling area <5% (cross-cut test)
  • Aging: Store at 65℃/75% humidity for 28 days (ASTM D573), simulating 2 years of use, Tritan material no embrittlement

Step Five:

After each testing round, list problem items, modify and retest:

• Problem 1: Handle pinches hand (5 people feedback)
  • Modification: Diameter increased from 3cm→3.8cm, use flexible TPU material (cost +$0.3/unit)
  • Effect: 20 people tested comfort score improved from 3.1→4.6
• Problem 2: Insufficient insulation (35℃ after 4h)
  • Modification: Vacuum layer thickness +0.5mm (cost +$0.3/unit), add reflective aluminum foil (reduces heat radiation)
  • Effect: Temperature 42℃ after 4h (meets target)
• Problem 3: Food residue in lid hinge
  • Modification: Open-style hinge (no dead corners), endoscope shows residue reduced 92%

Philadelphia company LunchSafe after 3 iterations, mass production failure rate dropped from 22%→3% (Packaging Digest 2023 data), total prototype investment per unit 1800 (3 rounds × 600), avoided $280,000/year customer complaint compensation.

Production

Production is the mass manufacturing stage to realize custom lunch box designs, using data to control quality: materials 100% pass FDA/LFGB/SGS certification, mold tolerance ±0.02mm (5-axis CNC machining), injection molding machine clamping force 80-500 tons (Engel/Husky models), AOI visual inspection yield 99.8%, final inspection includes 1.2-meter six-side drop (ASTM D4169), -20℃~120℃ temperature cycling (ISO 188), ensuring each lunch box is safe, consistent, and durable.

Safety and Performance

Material safety isn’t just talk, each certification needs hard data

LyondellBasell’s PP must have FDA 21 CFR 177.1520 report, proving total migration ≤10mg/dm² when microwaved at 120℃;

Eastman’s Tritan must have LFGB Annex I certification, test items include sensory evaluation (no off-odor), potassium permanganate consumption (≤10mg/kg), heavy metals (lead ≤0.01mg/L, cadmium ≤0.005mg/L).

Metal parts stricter, Outokumpu’s 304 stainless steel must pass SGS EN 10088-2 test, simulating acidic food (pH 3.5 lemon juice) immersion for 24 hours, heavy metal migration must be below EU EC 1935/2004 limits.

For example, plastic “total migration” per EU 10/2011 standard, uses 10% ethanol to simulate fatty foods, placed at 70℃ for 10 days, measures total dissolved substances;

Stainless steel “corrosion resistance” uses salt spray test (ASTM B117), 5% sodium chloride solution spray for 48 hours, surface must show no rust spots.

This data is directly printed on the material’s COA (Certificate of Analysis), factory incoming materials can scan to retrieve the original report.

Performance must be “tailored” to use, even a slight difference is unacceptable

For example, microwave-safe models must use PP, because PP melting point 165℃, withstands 120℃ without deformation, while Tritan although clear only withstands 109℃, microwaving causes bulging.

Insulated models more complex, inner pot uses 304 stainless steel (thermal conductivity 16W/m·K), middle vacuum insulation layer (thermal conductivity ≤0.03W/m·K), tests show holding 90℃ hot water, water temperature remains above 55℃ after 6 hours.

Clear models choosing Tritan isn’t just about light transmittance (92%, measured with X-Rite spectrophotometer), must also be impact-resistant.

Models with soup compartments, sealing gasket uses food-grade silicone (Loctite 518), tested per FDA 21 CFR 177.2600, boiled in steam at 121℃ for 30 minutes, volume change rate ≤5%, ensuring no leakage after repeated opening/closing.

Even the same batch of material must be identical, otherwise lunch boxes will “act up”

Factory uses X-Rite Ci64 spectrophotometer to measure color difference, same batch PP ΔE ≤1.5 (just noticeable color difference for human eye is ΔE=3), exceeding triggers return.

Thickness measured with ultrasonic thickness gauge (Olympus 38DL PLUS), PP box body nominal 1.2mm, actual measurement must be between 1.1~1.3mm, tolerance ±0.1mm.

Sample 10 pieces per batch, use Instron universal testing machine for Izod impact test (ASTM D256), pendulum energy 2.75J, energy absorbed at notch fracture must be ≥30J/m², ensures less likely to crack when dropped.

Last year a batch of Tritan material tested at 28J/m² impact absorption energy, directly returned to Eastman for replacement, fearing boxes made from it would crack under heavy contents.

Replicating the Design

First “break down” the design into language molds understand

When building 3D digital models in SolidWorks, engineers first specify draft angles for each structure—box body side walls 1.5°, deep cavity bottom 3°, ensures no surface damage during demolding.

Cooling channels are key, spiral water channels according to product shape, e.g., for a 100mm diameter round lunch box, water channel 5mm from cavity, can shorten cycle time from 40 seconds to 32 seconds (20% reduction).

Must also calculate shrinkage rates: PP shrinks 1.5%, Tritan shrinks 0.6%, when designing the digital model, dimensions are enlarged proportionally, e.g., design spec 100mm, PP mold cavity must be made 101.5mm.

Use high-precision CNC machines to “carve” the mold shape

Mold blank selects pre-hardened steel P20 (hardness HRC 30-35), uses 5-axis CNC machining.

Haas VF-4SS machine mills the mold core, spindle speed 12000rpm, feed rate 8000mm/min, critical contours (e.g., lid edges) tolerance controlled within ±0.02mm.

Cavity surface must “mirror-like”, first rough polish with #400 sandpaper, then fine polish with wool wheel + diamond paste to Ra ≤0.05μm (mirror effect);

Anti-slip patterns use DMG MORI laser etching machine, depth 0.1mm, spacing 0.5mm, feels slightly textured but not rough to touch.

Trial molding is like a “rehearsal”, adjust parameters until identical

Use Engel e-motion 310 injection molding machine, set melt temperature per design: PP 230℃, Tritan 250℃, injection pressure gradually increased from 80MPa to 150MPa, observe if cavity fills completely.

First shot parts measured with Zeiss CONTURA G2 coordinate measuring machine—box body length 200mm, allowed ±0.5mm, exceeding adjusts packing time (increase from 5 seconds to 8 seconds).

Run 50 consecutive shots, use air gauge to measure clasp hole diameter (φ6mm±0.1mm), statistical tolerance distribution, ensure over 95% within tolerance zone.

Trial molding focuses on three issues:

• Flash: Parting line overflow exceeding 0.1mm requires mold repair (use EDM electric discharge machine to remove excess material)
• Sink marks: At sudden wall thickness changes (e.g., box body to base transition) add cold slug wells, or increase cooling water temperature (from 20℃ to 30℃)
• Warpage: Warpage measured with flat plate + feeler gauge, lid arching ≤0.3mm passes

Not afraid of design changes, molds can “adapt flexibly”

For example, original logo depth 0.5mm, changed to 0.3mm, uses high-speed engraving machine (Roland MDX-540) to re-mill the original cavity, accuracy ±0.01mm.

Changing clasp curvature, remove insert for separate machining—use Wire EDM (Sodick AQ550L), cutting width 0.2mm, after cutting polish to Ra 0.1μm.

After each change, re-run trial molding, data updated into SOP file, e.g., “Mold revision 3, clasp pressure adjusted from 4N·m to 3.5N·m”.

Replicating multiple components together, must “align” to look good

Design uses a shared coordinate system, box body locating hole φ8mm±0.05mm, inner compartment plug φ7.9mm±0.05mm, after assembly gap ≤0.2mm (tested with 0.2mm feeler gauge doesn’t pass).

Use Cognex In-Sight 8400 vision system to capture assembly image, compare to design digital model, misalignment over 0.1mm triggers alarm.

During trial production of a three-layer lunch box, inner compartment was 0.3mm higher than box body, investigation found inner compartment mold shrinkage miscalculated (over-compensated 0.2%), after mold modification retest OK.

Replication accuracy, decided by data “verdict”

Each mold set passes three acceptance stages:

1. First Article Inspection (FAI): CMM measures 20 key dimensions (total length, clasp hole spacing, wall thickness), recorded in FAI report
2. Continuous Production Test: 8-hour run producing 1000 parts, automatic thickness gauge (Laselec ML 3000) monitors wall thickness, deviation >5% stops for parameter adjustment
3. Lifespan Test: Simulate mass production for 100,000 cycles, measure key dimensions every 20,000 cycles, wear exceeding 0.05mm requires mold repair

Modular Production Line

Forming Module:

Forming is the first station on the line, using high-speed injection/thermoforming machines + quick mold change systems.

For example, Husky HyPET HPP5 injection molding machine, clamping force 500 tons, with quick mold clamping system (Stäubli TP80), mold change: release 16 clamps, remove old mold, install new mold, clamp, total 12 minutes (including preheating).

Molding PP lunch boxes: melt temperature 230℃, injection pressure 120MPa, cycle 22 seconds/part, output 163 parts/hour; Tritan material slightly longer cycle (28 seconds/part), 128 parts/hour.

Thermoforming module uses ILLIG RV 74 machine, making thin-wall outer boxes (0.3mm thickness), sheet material (APET) heated to 160℃ softened, vacuum-formed, cycle 30 seconds/part, output 120 parts/hour.

Molds use standardized interfaces, e.g., box body mold and lid mold both use DME A-series mold bases, changing styles only requires changing mold core, not entire mold base, saves 2 hours.

Assembly Module:

Assembly module fears wrong part installation, uses FANUC robotic arm + Cognex vision to solve.

For example, three-layer lunch box (box body + inner divider + lid), robotic arm first grabs box body places on positioning station, vision camera (In-Sight 8400) captures clasp position, sends to control system;

Then grabs inner divider, camera compares divider plug and box body locating hole, deviation >0.1mm triggers alarm, robotic arm automatically fine-tunes.

Sealing gasket installation uses automatic press-fitting machine (DELO AD830), pressure set 5N±0.5N, press time 2 seconds, ensures silicone gasket doesn’t deform.

Clasps use ultrasonic welding (Branson 2000X), frequency 40kHz, power 800W, weld time 0.8 seconds, weld strength ≥15N (tensile test).

A model with soup compartment, after assembly misalignment rate <0.01% (1 error per 10,000 parts), relies on vision-guided repeat positioning accuracy ±0.05mm.

Inspection Module:

Online uses Omron FH-5050 AOI system, installed at end of assembly line, captures 30 images per second, identifies flash >0.1mm², short shots, assembly misalignment, e.g., missing a clasp triggers immediate stop.

Offline uses Zeiss CONTURA G2 CMM, samples 5 parts every 2 hours to measure key dimensions, data stored in MES system, out-of-tolerance automatically flagged red.

Sealing test uses automatic pressure tester (ATEQ F620), after filling with water and inverting, applies air pressure (50kPa), holds 30 seconds, leakage >0.1mL/min fails.

A factory tested 100,000 parts, sealing pass rate 99.9%, relies on pressure tester accuracy ±0.5kPa, 5 times faster than manual testing.

Packaging Module:

Packaging module connects to WMS system (Manhattan Associates), after receiving orders, system allocates SKUs to corresponding packaging stations.

For example, Amazon FBA orders, lunch box packed in EPE foam (10mm thickness), color box uses 350g white cardstock (FSC certified), LOGO UV printed.

Automatic weighing scale (Mettler Toledo BC-60) measures weight, deviation ±5g from standard triggers alarm (fear of missing accessories);

Automatic label applicator (Zebra ZT410) applies shipping label, QR code scan retrieves production batch, QC report.

Packed boxes enter automatic palletizer (KUKA KR QUANTEC), stacks 1.2m high pallets, WMS dispatches AGV (MiR 500) to warehouse, entire process unmanned.