The hum of machinery is a lot more than just background noise for PepsiCo. By using the latest sensor tech, it has become a powerful diagnostic tool.
In the evolving landscape of modern manufacturing, the integration of advanced technologies has become a pivotal element for companies to enhance efficiency and eliminate downtime. A prime example of this technological adoption is PepsiCo’s experimental use of AI-powered sensors in its production lines. The initiative underscores a broader industry trend towards digitalization and proactive maintenance strategies to ensure seamless operations.
“At Augury, we collaborate with leading manufacturers to enhance the reliability, productivity, and sustainability of their production lines and machinery,” said Augury CEO and co-founder Saar Yoskovitz. “Our unique approach is essentially ‘listening’ to machines to detect potential issues. To accomplish this, we’ve developed AI-powered sensors that monitor vibration, temperature, and magnetic emissions from motors. Analyzing changes in these factors allows us to predict malfunctions weeks or even months in advance.”
Yoskovitz explained that by utilizing sophisticated AI algorithms, the company can diagnose specific problems that would otherwise be indiscernible. More than 300 million hours of machine monitoring forms the basis for these algorithms, which are capable of identifying various malfunctions and flagging them for preemptive maintenance. This technology not only prevents downtime but plays a crucial role in optimizing machine efficiency and reducing waste.
At the heart of this transformation is the acknowledgment of an unconventional yet critical factor: noise. Contrary to the common perception that the clamor of the manufacturing machinery is just background noise, PepsiCo and Augury have turned it into a diagnostic tool.
As machinery wears down, it emits subtly different sounds. If accurately interpreted, can signal the need for maintenance before a breakdown occurs. This approach is particularly crucial in high-stakes environments like PepsiCo’s snack production facilities, where thousands of bags are produced daily, and any halt in the production line leads to significant losses and supply chain disruptions.
Augury’s technology employs sensors to monitor three key indicators of machine health: vibration, temperature, and magnetic emissions. By analyzing these parameters, the AI can detect potential malfunctions weeks or even months in advance, allowing for preemptive repairs. It’s not just about preventing downtime—it contributes to sustainability by optimizing machine performance, reducing energy consumption and waste.
The deployment of these sensors at PepsiCo marks a significant step towards the digitalization of the manufacturing sector. By leveraging real-time data and analytics, the company is not only able to maintain continuous production but also improve the quality of its products. The successful pilot in the U.S. has paved the way for a global rollout, indicating the potential for this technology to transform manufacturing practices worldwide.
The integration of Augury’s sensors and AI algorithms represents a fusion of hardware and software that provides actionable insights into machine health. As PepsiCo continues to explore and implement such solutions, its aim is towards more efficient, reliable, and sustainable production methods that could redefine industry standards and operational excellence.
For Augury’s business model, sound and vibration are one in the same. Understanding the intricacies of machine health is crucial for operational efficiency and longevity. Vibration analysis is a cornerstone method that allows teams to detect early signs of equipment failure, thereby extending operational hours and enhancing overall equipment performance.
Data collection methods
There are two primary methods of data collection in vibration analysis: route-based processes and online monitoring solutions. The route-based approach offers a cost-effective entry point, enabling teams to build maintenance routes with simple tools and local support. This method is not only budget-friendly but also an opportunity for skill development within maintenance teams. However, its drawbacks include the potential for significant delays between data collection and analysis, given the infrequency of third-party consultant visits and the time constraints on personnel who often juggle multiple responsibilities.
On the flip side, online monitoring solutions provide continuous, real-time data collection, eliminating the time lag associated with route-based methods. This approach ensures immediate analysis and feedback on equipment health, allowing for swift interventions if necessary. It’s scalable across sites and facilitates the sharing of best practices among different locations within a company. Despite its advantages, the initial setup can be costly and the absence of local support may pose challenges for some organizations.
What’s that sound?
A rotating shaft produces vibrations that can be visualized as a sine wave, with variations in wave frequency indicating different shaft speeds. Real-world applications, however, involve complex signals composed of multiple frequencies generated by various machine components. Through a technique known as Fast Fourier Transform (FFT), analysts deconstruct these complex signals into individual sine waves, providing a clear picture of machine health by identifying specific fault frequencies.
Vibration trends over time could signal increasing wear or failure in machine components like bearings. By labeling these frequencies on a spectrum, analysts can pinpoint the source of the problem, such as lubrication issues or bearing wear, facilitating targeted maintenance actions.
Vibration analysis is a critical element of modern maintenance strategies, offering a blend of traditional techniques and the latest technology to ensure machinery operates at peak efficiency. As manufacturers navigate their maintenance needs, adopting and integrating such predictive tools could become essential in minimizing downtime and advancing operational success.
There are myriad fault types that can significantly impact machinery performance. Understanding these fault types through vibration analysis aids in maintaining operational efficiency and also in extends the lifespan of critical machinery components.
One of the primary faults vibration analysis can detect is bearing wear and lubrication issues, characterized by non-synchronous frequencies that don’t align with whole number intervals of the shaft speed. This discrepancy allows analysts to gauge the severity of bearing wear over time, providing valuable insights into the maintenance needs of the machinery. As bearing wear progresses, the frequencies detected move from higher to lower regions, signaling the development of the wear and the need for intervention.
Coupling wear and misalignment represent another set of challenges that vibration analysis can pinpoint. Symptoms of these issues include high axial vibration and radial vibration at two or three times the shaft speed. This can escalate to a resemblance of rotating mechanical looseness as the wear worsens, illustrating the interconnected nature of machine faults.
Gear wear, another critical fault type, manifests through gear mesh frequency, determined by the number of teeth on the shaft multiplied by the shaft speed. The complexity of gearboxes, with their multiple shafts and sets of teeth, introduces a variety of gear mesh frequencies and secondary frequencies that help in diagnosing specific gear wear faults, such as broken teeth or misalignment.
Sheave issues and belt wear also present unique diagnostic challenges. Vibration analysis helps identify the speeds of different shafts involved, including the belt pass frequency, which is indicative of belt health. High vibration in line with the belts can point towards eccentric sheaves or imbalance, while high axial vibration suggests misalignment. Additionally, an elevated belt pass frequency could indicate wear or improper tension, emphasizing the critical role of accurate frequency identification in troubleshooting.
Finally, structural and rotating looseness, although similarly named, are very distinct fault types. Structural looseness (Type A) is related to the foundation or mounting of the machine and is typically indicated by a significant disparity in vibration across different directions. Rotating mechanical looseness (Type C), on the other hand, occurs between stationary and moving parts, like a loose bearing, producing harmonics or whole number multiples of the shaft speed.
Real-time precision
This nuanced understanding of fault types provided by Augury’s technology is essential for effective maintenance strategies, allowing for precise diagnosis and timely intervention to prevent costly downtimes and extend equipment life.
“We scour the globe for cutting-edge technologies, aiming for swift piloting to facilitate quick decision-making on whether to proceed or halt further deployment,” said PepsiCo Labs Global VP of Tech Venturing and Innovation David Schwartz. “If a solution demonstrates its efficacy, we’re committed to scaling it internationally. What excites us particularly is how it merges two innovative technologies: sensors that allow us to monitor our factory operations in real-time, and artificial intelligence, which synthesizes this data into actionable insights.”
