RESEARCH REPORT · Technical Research AUREK-RC-AR-038

Reliability of Vacuum Gripping and Dynamic Rotation of Curved Automotive Glass

Study sealing formation, all-orientation suction loads, dynamic tilting moments, vacuum zoning and reserve, surface and edge protection, fault injection, and validation using production-representative curved automotive glass samples.

Key points of this article
  • Local curvature matching, sealing lip preload and local normal alignment are the prerequisites for reliable vacuum gripping of curved glass.
  • Normal, tangential, and eccentric moments continue to change during flipping, and the full angular trajectory must be verified rather than just looking at the endpoints.
  • The holding time should be determined from the true leakage curve under the most unfavorable surface and wear conditions.
  • Low-mark materials still require sample re-inspection under specified temperatures, cleaning and contact times.
Judgment rules:Document Properties: Public Information Research and Engineering Verification Methods. This article does not claim that any specific fixture, glass model or production line has passed the test; the example parameters and criteria need to be re-confirmed by real samples and project risk assessment.

00executive summary

Windshields, rear windows, side windows, and panoramic canopies have airtight surfaces that are generally suitable for vacuum handling, but their curvature, laminate construction, silkscreen/coating, edge fragility, and dynamic flipping make reliability issues significantly different from flat glass. Whether suction cup can form a stable seal depends not only on the pressure difference and area, but also on the contact geometry, preload, bending and sealing lip load distribution between suction cup and the local curved surface. IEEE's suction cup modeling study separated axial compression and bending force, and verified the load-bearing prediction idea through curved surface tests [1]; industrial data emphasizes the role of flexible sealing lips, corrugation compensation, limited support surfaces and low-imprint materials in automotive glass [2][3].

During dynamic rotation, load is continuously redistributed among normal force, tangential force, and moment. Tooling that maintains stable vacuum in the initial vertical orientation may experience its maximum rotation torque at the intermediate 45° orientation because of combined center-of-gravity offset, or may fail in the horizontal orientation because glass deflection redistributes suction-point loads. Engineering design shall therefore treat seal establishment, holding, full-angle rotation, positioning, support-confirmed release, and abnormal controlled set-down as one continuous state machine.

Core conclusion

  • The number and total area of the suction cups are only the starting point. Matching the local curvature, correctly preloading the sealing lip, and aligning each suction-cup center axis with the local surface normal are prerequisites for reliable suction.
  • Dynamic flipping should calculate the gravity component, eccentric moment, inertia moment and braking demand according to the attitude angle. During verification, vacuum, angle, acceleration and relative slip should be continuously collected.
  • The vacuum system shall use appropriate zoning, check valves/energy storage, inhibition of motion at low vacuum, and support-confirmed release. Pressure-retention time shall be derived from leakage curves rather than an empirical constant.
  • Glass protection needs to simultaneously control edge impacts, metal proximity, suction cup marks, chemical fingerprints and cleaning particles.
  • Reliability release should cover the combination of curvature limit, surface condition, temperature, protective film, accessories, incoming material deviation and suction cup wear.

01Research objects and applicable boundaries

The research object is automotive safety glass and its pre-assembly assembly, including laminated windshields, tempered side windows/rear windows, glass with silk screen, coating, heating wire, wrapping, positioning pins, wiring harness or camera bracket. GB 9656-2021 provides safety technical specifications for motor vehicle glass [4], and ISO 3537:2015 specifies mechanical test methods for safety glass materials [5]. These product standards do not directly give handling tooling parameters, but remind designers: glass is a product defined by safety performance, and tooling should not introduce unassessed scratches, chipping, local damage or coating effects.

This report covers vacuum tooling on manually guided Pneumatic Industrial Manipulators, robot end effectors, and purpose-built rotation mechanisms. It excludes forming in glass-manufacturing furnaces, complete-vehicle crash performance, and adhesive-curing processes. Customer requirements for appearance, optics, or coatings shall be governed by the customer's latest technical specifications.

01 Research objects and applicable boundaries — data table
Data categoryRequired inputImpact on tooling design
glass geometryThree-dimensional surface, thickness, dimensions, local curvature, edge profilesuction cup type, local normal, layout and envelope
Quality attributesGlass mass, accessory mass, combined center of gravity, inertiaLifting capacity, tilting axis and braking torque
surface systemSilk screen printing, coating, protective film, ink, cleanerPermitted suction zones, material compatibility, and marking
Assembly accessoriesEdge encapsulation, pins, wiring harnesses, brackets, adhesive-bead pathsRestricted area, tooling avoidance and release path
Incoming material statusRack angle, pane spacing, adhered panes, placement variationApproach sequence, suction confirmation, and pane-separation motion

02Curved surface sealing and vacuum gripping load bearing model

02.1 Conditions for seal formation

The first question in suction on a curved surface is not how much weight it can carry, but whether the sealing lip can maintain continuous contact under limited preload. Excessive misalignment between the suction-cup center axis and the local surface normal over-compresses one side of the sealing lip and lifts the other. Bellows and flexible connections can accommodate angular and height differences, but they also increase orientation compliance and tangential displacement. Suction-cup model studies show that contact geometry, axial compression, and bending deformation jointly determine sealing and load capacity [1].

In engineering, the following constraints can be used to filter suction points:

  • Within the effective sealing diameter range of suction cup, the glass curvature is compatible with the allowable curvature of suction cup.
  • The suction-cup preload displacement shall remain within the supplier-recommended range; do not force the manipulator against the glass to compensate for incorrect alignment.
  • The sealing lip completely avoids silk screen steps, protective film folds, holes and accessory boundaries.
  • The flexible support stroke can absorb the incoming material deviation, but will not lose sufficient anti-swing stability once it is in place.

02.2 Normal, tangential and combined loads

F_N,usable = (Δp_min · ΣA_eff · η_curve · η_surface) / S_N

Here, η_curve is the reduction factor for curvature and orientation, and η_surface is the reduction factor for surface condition, contamination, temperature, and wear. Gravity on vertical glass creates primarily a tangential demand; during rotation, the normal and tangential components vary with angle. If θ is defined as the angle between the glass normal and the vertical direction, resolve gravity and inertia in the project coordinate system and calculate suction-cup load distribution at every discrete orientation, rather than checking only the 0° and 90° endpoints.

F_T,demand(θ) = m · |g_T(θ) + a_T(θ)| + F_route(θ) M_demand(θ) = mge · sin(θ) + Iα + cω + M_disturbance

In the formula, e is the eccentricity from the center of gravity of the glass-tooling combination to the flip axis, I is the moment of inertia, α and ω are the angular acceleration and angular velocity, and cω represents the equivalent damping term. The maximum torque does not necessarily occur at the end point and must be checked with the complete action curve.

03Suction cup layout and vacuum structure

The suction cups shall form a stable support polygon around the combined center of gravity while avoiding edges, black ceramic-print steps, heater-connection areas, camera/sensor brackets, edge encapsulation, and adhesive-bead paths. For large glass, control cup spacing so that local deflection and center sag remain within product limits. The layout shall also preserve the operator's view during installation: the tooling shall not obscure critical locating points or enter collision zones around the A-pillars, headliner, or instrument panel as it approaches the body.

03.1 Vacuum partition

The objective of vacuum zoning is to limit propagation of a single-point leak and improve diagnosability, not simply to make every suction cup independent. Each zone shall have sufficient residual capacity, suitable check valves, and a measurable pressure point. Sensor locations shall represent the state at the suction-cup side rather than only the vacuum-source outlet. Excessive zoning adds valves, lines, and leak points, so reliability and maintainability must be balanced.

03.1 Vacuum partition — data table
HierarchyDesign pointstarget state at failure
vacuum generationCan cover establishment time and known leaks, with operational monitoringDo not permit a new lifting cycle when source capacity is insufficient
Vacuum Zoning and Check ValvesA single zone failure does not immediately bring down all suction cupMaintain, stop dangerous actions, and alarm
Energy storage/volumeProvides time for controlled placement and does not replace mechanical safety measuresPressure decay is predictable and diagnosable
threshold monitoringSeparate establishment thresholds, operational thresholds and alert thresholdsLimit flip/release when threshold is crossed
release loopVent only when the glass is supported and the orientation permits releaseRefuse to release if any condition is not met

03.2 Pressure holding and leakage curve

The ideal-gas equation cannot fully describe actual leakage as sealing-lip contact changes with load. Measure the Δp(t) curve on actual glass and determine the time window from the operating threshold to the minimum holding threshold under worst-case surface condition, temperature, suction-cup wear, and orientation. If a local interval approximates exponential decay, fit the time constant τ using the following equation, but base acceptance on the envelope of the measured curves.

Δp(t) ≈ Δp₀ · exp(-t/τ)

04Dynamic flipping and operator control

The flip axis should be as close as possible to the center of gravity of the combination to reduce operator force and braking burden. Complete center-of-gravity balancing is not always optimal: a controlled sense of direction needs to be retained while preventing the glass from flipping over on its own. Rotating mechanisms should have verifiable braking/locking capabilities and controlled release, and hoses and cables must be free from pulling, bending, and snagging at all angles.

04.1 Example torque (only method explained)

Assume the combined mass of the glass and tooling is 40 kg and the combined center of gravity is offset 0.08 m from the rotation axis; the peak gravitational eccentric torque is approximately 31.4 N·m. If maximum inertial torque is estimated at 8 N·m, and friction variation and external disturbances are also considered, the brake, lock, and permitted operator force shall be designed with additional margin based on the risk assessment. The values 40 kg, 0.08 m, and 8 N·m are example assumptions and are not ratings for any product.

The dynamic test should be set up in two layers: slow exploration trajectory and production trajectory. First confirm seals, envelopes and pipelines at low energy, then gradually increase speed and acceleration; only change one main variable at each change, and retain the data version. Trial and error directly at production speed amplifies the risk of glass damage and falls.

05Glass surface, edge and accessory protection

Schmalz's automotive glass data points out that suction cup marks are related to the structure, material and relative movement of the sealing lip on the glass surface, and contamination particles will amplify the risk of marks [2]. Therefore, "low-marking materials" are not exempt from inspection. Sample validation should be completed for each glass, surface, cleaning agent, contact time and temperature combination.

05 Glass surface, edge and accessory protection — data table
riskmechanismcontrol measuresCheck method
suction cup imprintMaterial migration, slippage, local surface pressureSpecify material, limit slip and contact timeCheck under specified lighting/angle
scratchGrit, metal proximity, dirty glovesCleanliness classification, soft separation, and enclosed storageSurface magnification inspection and defect map
Edge chipping/micro cracksEdge impact, snagging on a rack edge, tooling interferenceSafe edge clearance, guidance, and low-speed removal from the rackVisual, dedicated inspection or customer regulations
Attachment forceThe wiring harness/bracket is caught by toolingThree-dimensional prohibited/restricted zones and path checkFull-orientation envelope validation
Coating/silk screen effectChemical incompatibility or unstable step sealPermitted suction zone and material-compatibility testProcess approval and appearance/function retest

There is also a risk of adhered panes and sudden separation when removing glass from a rack. The tooling must confirm that a single pane has separated from the rack and, where necessary, use a controlled short stroke, inter-pane sensing, or an auxiliary separation method. Do not rely on rapid pull-off to separate adhered panes.

06Verification Matrix and Fault Injection

06 Verification Matrix and Fault Injection — data table
Verification scenariovariable combinationRecord signalMain criteria
curvature limitMinimum/maximum curvature, incoming material deviationVacuum and preload displacement of each zoneAll key areas are stably established
surface boundaryCleaning, specified contamination, protective film, temperatureVacuum-establishment time, pressure decay, markingNo threshold exceedance and acceptable surface condition
Full-angle rotationContinuous trajectory from 0° to target angleAngle, velocity, acceleration, vacuum, slipNo peeling, slipping or uncontrolled rotation
Emergency stop/brakeThe most unfavorable angle and speedStopping time, peak acceleration, torqueNo falling, no collision, recoverable
single zone leakIntroduce a controlled leak or isolate one suction cupZone pressure, control statusStop hazardous motion while retaining sufficient time for controlled set-down
Release command by mistakeGlass suspended / not supportedRelease valve and status feedbackThe command is rejected and an alarm is generated.
Retest after durabilityRepresentative cycles and cleaning cyclesLeakage trends, material wear, marksMaintenance thresholds can be detected and performance can be maintained

Fault injection shall begin with low energy, surrogate loads, and simulated signals; actual glass shall be used only at the final stage under controlled-risk conditions. Each test shall define the injection location and time, expected system state, permitted response window, recovery conditions, and evidence file. Fault injection is not intended to prove that failure is impossible; it confirms that a single fault or foreseeable combination of faults does not directly cause a hazardous release.

07FMEA and control state machine

07 FMEA and control state machine — data table
failure modeConsequencesDetection/ControlSafe recovery
suction cup does not fit completelySlow leak or peeling after liftingPartition establishment time and pressure trendDo not permit departure from the rack; re-establish suction
Single zone pipeline ruptureVacuum drops rapidlyCheck valve, per-zone sensing, and threshold interlockStop rotation and perform a controlled set-down
Angle sensing abnormalityWrong brake/release permissionRationality check, mechanical limitLockout and manual diagnosis
Brake is not lockedFlip and collide on its ownLocking feedback, torque/position abnormalityLimited movement, safe support
Wrong glass modelSuction points enter restricted areas or path interferenceVehicle model–tooling–recipe consistencyInhibit suction and lifting
Release valve stickingAccidental air intakeValve status diagnosis, dual condition releaseGeometric support / per-zone pressure retention
suction cup contamination wearMarks, leaks, scratchesCleaning inspection, life and trend monitoringFirst item confirmed after replacement

The recommended state machine includes: unloaded safe state, approach, preload, vacuum establishment, suction confirmation, departure from the rack, transport, rotation, positioning, support confirmation, release permission, release complete, and abnormal-load retention. Any sensor conflict, communication interruption, or threshold exceedance shall place the system in the conservative state: retain the load, stop hazardous motion, and inhibit release. Recovery shall first clear the fault and confirm that the glass is in a supported position.

08Implementation information and conclusions

Engineering deliverables shall include at least the glass-family parameter matrix, permitted suction zones, local surface normals and suction-cup installation drawing, vacuum zoning and thresholds, rotation-axis and torque calculations, trajectory envelope, fault-state table, raw data from actual-part tests, surface-inspection rules, maintenance intervals, and abnormal set-down instructions. At multi-model mixed-production stations, interlock the glass model, tooling module, and control recipe for consistency.

Reliable vacuum handling of curved automotive glass depends on four closed loops: a genuine geometric seal, adequate load margin at every orientation, monitored control without inadvertent release, and preservation of surface and edge quality. No single measure can replace complete verification. This report may serve as the basis for design reviews and test plans, but it is not product certification, customer acceptance, or a third-party test report.

References

  1. Hudoklin, J. et al. Vacuum Suction Cup Modeling for Evaluation of Sealing and Real-Time Simulation. IEEE Robotics and Automation Letters, 2022. https://doi.org/10.1109/LRA.2022.3145509
  2. J. Schmalz GmbH. Vacuum-Handling Systems / Suction Cups for Automotive Glass Handling. https://www.schmalz.com/site/binaries/content/assets/media/04_applications/industries/automotive/en/Vakuum-Handhabungssysteme_AMI_HS_2020_englisch.pdf
  3. J. Schmalz GmbH. Bellows suction cup SAB HT1 for dynamic handling of automotive glass. https://www.schmalz.com/en-us/products/vacuum-technology-for-automation-301607/vacuum-components-301608/vacuum-suction-cups-301609/suction-cups-for-handling-glass-305667/bellows-suction-cups-sab-ht1-15-folds-305668
  4. GB 9656-2021 Safety technical specifications for motor vehicle glass. https://openstd.samr.gov.cn/bzgk/std/newGbInfo?hcno=9D70607A5C909E556DC2D140CE56FE3A
  5. ISO 3537:2015. Road vehicles - Safety glazing materials - Mechanical tests. https://www.iso.org/standard/56215.html
  6. Novotny, F.; Horak, M. Computer modelling of suction cups used for window cleaning robot and automatic handling of glass sheets. MM Science Journal, 2009. https://doi.org/10.17973/MMSJ.2009_06_20090304
  7. ISO 12100:2010. Safety of machinery - General principles for design - Risk assessment and risk reduction. https://www.iso.org/standard/51528.html
  8. ISO 4414:2010. Pneumatic fluid power - General rules and safety requirements for systems and their components. https://www.iso.org/standard/44790.html
  9. ISO/TR 20218-1:2018. Robotics - Safety design for industrial robot systems - Part 1: End-effectors. https://www.iso.org/standard/69488.html
Need a car moving station plan evaluation?Please provide the workpiece weight, dimensions, on-site layout, complete movements and rhythm, and AUREK can assist in checking the equipment and fixture solutions.
View automotive industry solutions

Frequently Asked Questions · FAQ

Can flat glass suction cups be used directly on windshields?

It cannot be directly inferred. There are double curvatures, silkscreens, attachments, and local normal differences in the windshield, and it is necessary to confirm that the suction cup allows for curvature, preloading, and placement.

Is the most dangerous flip position necessarily 90°?

Not necessarily. Maximum eccentric moments, tangential loads or suction point redistributions may occur at intermediate angles and should be calculated and tested for the complete trajectory.

Can an air receiver guarantee that the load will never drop?

No. Energy storage can only delay vacuum decay, and still requires zoning, check, status monitoring, action interlocking and controlled placement.

How to verify that the suction cup will not damage the coating or silk screen?

Recheck to customer appearance and functionality standards using real glass, specified cleaning agents, temperature, contact time and number of repetitions.

Does the report give the number of universal suction cups?

Not given. The quantity, diameter, threshold and holding time must be determined in conjunction with glass mass, curvature, attitude, attachment and movement conditions.

Leave your moving problems to us Assessment

Whether the application involves retrofitting an existing line, supporting multiple models, avoiding welding equipment, or handling battery packs, tires, or automotive glass, confirm the solution using the actual workpiece and site conditions.