introduction
Project Kuiper is Amazon’s low earth orbit (LEO) satellite broadband network. The network is intended to provide fast, affordable connectivity to communities around the world that are unavailable or underserved by traditional internet and communications options. The network also has the performance, capacity, and flexibility to serve a wide range of business, communications, and government customers. To achieve this goal, Amazon is deploying thousands of satellites in low earth orbit (LEO) linked to a global network of ground-based antennas, fiber, and internet connection points.
High-tech manufacturing, including advanced CNC (Computer Numerical Control) machining, produces precision components for Project Kuiper’s broadband satellites. Leveraging cutting-edge technology throughout the manufacturing process, the Project Kuiper team will assemble and integrate these components into thousands of satellites. To optimize these complex manufacturing operations, they needed a solution to capture, organize, calculate, analyze, and monitor critical measurements for near real-time (NRT) monitoring and instrumentation analysis.
The team decided to build their solution using AWS IoT SiteWise, a managed service that collects, stores, organizes, and monitors industrial equipment data at scale. By leveraging AWS IoT SiteWise’s industrial data modeling and processing capabilities, Project Kuiper was able to generate data-driven insights that improved operational and performance efficiency, and production quality. In this blog, we discuss the challenges Project Kuiper faced in their operations, the solution architecture they deployed, and the business impact they achieved.
Opportunity | Improving operational efficiency with AWS IoT SiteWise
Project Kuiper’s high-tech manufacturing process uses CNC machines to transform raw materials such as aluminum and composites into intricate parts and components for the broadband satellites. These components include antenna reflectors, mounting brackets and machine housings, all of which require complex geometries and tight tolerances. Automated milling, turning and grinding operations enabled by CNC machines ensure consistent quality throughout mass production, enabling the team to maintain the precision and reliability required for Project Kuiper’s cutting-edge satellite technology.
Because parts were highly customized and design changes were frequent, it was important for the team to have NRT collect and analyze manufacturing data. This AWS IoT SiteWise analysis enabled them to quickly adjust their manufacturing processes. These adjustments minimized machine downtime, defects, and waste, and maximized quality and efficiency. However, collecting data to give NRT visibility into production key performance indicators (KPIs) and equipment performance was difficult. To achieve this, the team tracked metrics such as Overall Equipment Effectiveness (OEE), an industry-standard metric that measures how well manufacturing time is being utilized to produce good quality parts. Monitoring OEE gave the team deep insights into loss categories, allowing them to identify operational bottlenecks and opportunities for improvement.
Solution | Enabling NRT Data Delivery for Proactive and Early Identification of Issues
To address these challenges, the Project Kuiper team implemented a solution leveraging multiple AWS services that helped them collect manufacturing operations data, calculate KPIs, monitor NRT dashboards, and perform long-term trend analysis.
AWS IoT SiteWise Edge software securely collected manufacturing data from CNC machines. It selectively forwarded this equipment data, or process data, to AWS IoT SiteWise in the cloud. AWS IoT SiteWise offers two ingestion mechanisms: a streaming ingestion API to ingest telemetry data within milliseconds, and a buffer ingestion API to process analytical data streams in batches. By leveraging both ingestion methods, Project Kuiper was able to compose a cost-effective, scalable data pipeline that supported its NRT monitoring and data analytics needs. This approach allowed them to optimize costs by sending only the data required for NRT monitoring over a streaming path and using the more cost-effective buffer ingestion for analytics applications.
AWS IoT SiteWise helped contextualize manufacturing data by creating virtual representations of physical assets, such as CNC machines, organized in a hierarchical structure. This contextualization allowed the Project Kuiper team to associate data streams (such as sensor readings, machine health, and performance metrics) with specific assets. Contextualized data is more accessible and easier to interpret for many stakeholders (such as engineers, operators, and managers) to quickly find, identify, and analyze relevant information.
The Project Kuiper team leveraged AWS IoT SiteWise multi-tiered storage for cost optimization, data lifecycle management, and system performance as their manufacturing operations expanded. The team defined data retention periods to keep the most recent and frequently accessed data in hot storage for real-time monitoring. AWS IoT SiteWise automatically moved older, less frequently accessed data to cost-efficient warm and cold storage tiers. This storage lifecycle strategy enabled them to retain historical data for longer periods for trends and insights, while ensuring fast query performance for real-time monitoring and analysis. The scalable storage solution accommodated Project Kuiper’s changing requirements as their manufacturing operations expanded and data volumes grew, without excessive costs or performance issues.
Data visualization plays a key role in monitoring the operational efficiency of the manufacturing process. AWS IoT SiteWise Monitor is used for the NRT operational dashboard. The Project Kuiper team used NRT runtime charts (mostly line graphs) to quickly identify abnormal conditions and escalate issues to help resolve them quickly. Engineering then drilled down into the impacted data points to understand their impact on other operational conditions. Dashboard users could also search for assets and properties to monitor and drag them into data widgets such as XY plots, timelines, and tables. The NRT dashboard tracked key metrics such as OEE, defect rates, cycle times, and overall throughput efficiency. For longer-term analysis and business intelligence, Project Kuiper leveraged Amazon QuickSight. QuickSight provided a wide range of capabilities to create management reports and conduct detailed data inspection over long historical periods.
Figure 1: High-level architecture
Results | Improved data-driven decision making optimized operational efficiency, quality and costs
Project Kuiper proved successful with NRT’s implementation of an AWS IoT SiteWise-based architecture to monitor and analyze CNC machine data. Leveraging AWS IoT SiteWise, SiteWise Monitor, and other AWS analytics tools, such as Amazon Athena and Amazon QuickSight, the company gained deep visibility into its manufacturing processes. Contextualized insights enabled data-driven decisions to optimize production efficiency, quality, and costs.
Since implementing the solution, Project Kuiper has seen improved OEE, reduced unplanned downtime, and increased asset utilization. The ability of NRT to detect and address quality issues has reduced scrap and rework, resulting in significant cost savings. Additionally, insights gained from historical data analysis have helped identify production bottlenecks and implement targeted process improvements, resulting in increased overall throughput.
“As an engineering leader, I’m excited about the value my team has gained from implementing AWS IoT SiteWise for near real-time manufacturing analytics. With intuitive cloud dashboards to evaluate effectiveness, quality, output, and downtime rates, we are now able to make data-driven decisions across our facilities.”
– Paul Palcisco, Production Director, Kuiper Production Operations
“We are excited to be part of Project Kuiper and proud of the operational efficiency gains the team has achieved by adopting AWS IoT SiteWise, especially the ability to monitor KPIs across their assets in near real-time. Dynamic data collection and dashboards to calculate operational equipment effectiveness (OEE), defect rates, cycle times, and overall throughput efficiency have given Project Kuiper greater insight into bottlenecks and how to resolve them.”
-Michael MacKenzie, General Manager, Industrial IoT and Edge, AWS
Conclusion
In this post, we described how Project Kuiper uses AWS IoT SiteWise to collect, store, organize, and monitor data from their manufacturing processes. The solution enabled the Project Kuiper team to identify inconsistencies, detect anomalies, and make proactive, data-driven decisions to optimize production efficiency and quality. Project Kuiper’s AWS IoT SiteWise work demonstrates the transformative power of NRT monitoring and data-driven decision-making in high-tech manufacturing.
learn more
Learn more about Amazon’s Project Kuiper initiative here, and to get started with AWS IoT SiteWise, see our developer guide.
About the Author
Avik Ghosh
Avik is a Senior Product Manager in the AWS Industrial IoT team, focusing on AWS IoT SiteWise service. With 18+ years of experience in technology innovation and product delivery, he specializes in Industrial IoT, MES, Historian, and large scale Industry 4.0 solutions. Avik contributes to the conceptualization, research, definition, and validation of AWS IoT service offerings.
Mani Nazari
Mani is an experienced systems and development engineer with deep expertise in manufacturing, aerospace, distributed systems and embedded technologies. He currently works at Amazon’s Project Kuiper as a systems development engineer for ground support equipment and space mecha assembly. Mani has over 10 years of experience in software engineering, factory automation and quality control. Prior to Amazon, he held engineering/leadership roles at Boeing where he was involved in developing factory automation systems and designing APIs.
Joyson Neville Lewis
Joyson is a Senior IoT Data Architect with AWS Professional Services. Before moving into Conversational AI and Industrial IoT, Joyson worked as a Software/Data Engineer. He helps AWS customers realize their AI vision with Voice Assistants/Chatbots and IoT solutions.
