The need to focus more attention on EV battery manufacturing quality has become increasingly important with the growing popularity of electric and hybrid-electric vehicles.
Mass production of lithium-ion batteries presents manufacturers with a variety of challenges. In fact, many new processes are being employed on a trial-and-error basis. But battery producers and their OEM customers are still focused on finding solutions for providing the longest driving range possible while avoiding quality issues and malfunctions that could lead to costly recalls.
Properly sealing lithium-ion battery cases and covers is critical to overall battery performance, safety and quality. Automated dispensing applications for batteries must be precise to achieve full performance and to avoid dangerous moisture and gas leaks.
Today, various methods are used to seal battery cases and covers, including polyurethane foam-in-place gasketing, tall urethane beads and self-expanding foam.
Battery pouches also need to be joined through the application of hot-melt pressure-sensitive adhesives or water-based adhesives. The adhesive is sprayed onto the battery pouches and the pouches are joined to make up a battery module.
A major challenge for these applications is to ensure that the hot-melt adhesive has actually been applied to the pouch, as it is difficult to verify that the clear, thin glue has been properly placed.
Part-to-part variation is a concern for most components in automotive manufacturing; this is especially true for batteries. In so-called “skateboard” designs, battery packs are packaged in the floorpan between the vehicle’s axles, which often results in waviness and imperfections in the large battery covers.
However, cover sealant without any gaps are required to assure overall quality and avoid safety concerns. Due to cycle time, it isn’t possible to reprogram dispensing robots for individual parts to accommodate variations of this type.
Overcoming part-to-part variation and applying sealant to specification has proven to be a major challenge that requires in-line feedback and real-time nozzle adjustments as sealant beads are dispensed.
Another concern is how to consistently dispense the right volume at the right location, whether via a cover sealant or TIM. Material viscosity, barrel changes, fluctuations in humidity throughout the day and robot speed changes all affect actual dispensed volume. To account for these factors, closed-loop feedback with the dispenser, robot and some type of in-line vision system is needed.
As with all robotics, automated dispensing relies solely on programming to complete a task. To overcome battery manufacturing challenges, “vision” is required for the robot and dispenser to perform optimally.
Some dispensers indicate how much adhesive or sealant is dispensed but can’t determine if or where the material has been applied. For example, a dispenser and robot may run a programmed path and dispense sealant whether a part is present or not. If the part is moved or fixtured incorrectly, the material may be dispensed in an incorrect location or on the plant floor.
Machine vision provides data to evaluate how the dispensing process is performing. It allows engineers to view and record whether or not dispensed adhesives or sealants have been properly placed on a part. Machine vision also ensures that the dispensed material meets width, height and volume specifications set by the manufacturer.
Inspection systems are required to control the dispensing process for battery manufacturing. Vision provides data for part traceability, which is vital for OEMs concerned with overall quality, warranty issues and potential recalls.
There are a variety of types and use cases for vision systems, including 2D and 3D technologies, as will as post-inspection and in-line applications. A major downside with 2D vision is that volume isn’t assessed, which is a crucial measurement in the dispensing process. Many 2D vision systems are post-inspection, which means they only start after sealant has been applied, and they require additional floor space—a precious manufacturing commodity—for lighting rigs, cameras, fixtures and other automation equipment.
In addition, 2D vision is unable to decipher the difference between a part and the adhesive bead when the color of the part and adhesive material are the same (black on black or gray on gray) because cameras can’t distinguish the contrast between the two.
There are many advantages to using a robust in-line 3D inspection system. When inspection is completed in-line there is no need for additional cycle time, floor space or more fixturing. With 3D vision, volume is measured and material color and part color can be the same without affecting the system’s ability to provide accurate measurements.
Another advantage: Without a 3D image of the bead, it would be impossible to recognize if it failed to meet center height and volume specifications because the nozzle has plowed right through the center. Finally, the biggest in-line 3D vision benefit is its ability to provide real-time feedback to the robot and dispenser for process control.
Vision systems can provide a plethora of data such as volume, height, width and location of dispensed beads on every dispensed part. This is great information, but only if it can be utilized.
To this end, data analytics programs can provide a clear visualization of dispensing performance and enable continuous improvements in efficiency and quality. Users can even pinpoint certain days or times when production issues occur and monitor performance in real time. Data analytics of trends and variations in-line can also provide necessary details to schedule preventative maintenance.
It’s also clear that dispensing processes can benefit greatly from robotic vision systems. The challenge is how to best improve the process utilizing the data provided from the vision system.
Machine learning, process control and artificial intelligence (A.I.) all can help improve the battery-dispensing processes. Utilizing data collected by machine vision, A.I. can be implemented to adjust the processes to meet specifications on every dispensing cycle. For example, if a bubble or gap is missed on the original cycle, A.I. can take control of the robot and dispenser to fill the gap before moving on to the next part, which yields significant savings in labor and scrappage costs.
When real-time feedback is provided, adjustments can also be made by A.I. software to overcome part-to-part variation. Vision can detect the waviness of a part not programmed in the original robot path. The software can then calculate and send the Z and X offset to the robot and adjust the path on the fly. This avoids nozzle crashes and dispenses a precision bead in the correct location.
Cost and material savings can also be achieved through in-line process control. Sealants, TIMs and adhesives can be expensive, running as much as $180 per gallon. Assuring specified volumes are dispensed on battery components is critical to keeping programs within budget.
Machine vision can inspect the volume of material dispensed on a part, making it possible for software to learn and adjust flow commands on the next cycle. Coherix 3D Volume Adaptive Control uses this method in battery manufacturing to dispense up to 40% less material, while still meeting volume specifications.
Connect With Us