The Internet of Things on AWS – Official Blog

The Blueprint for Industrial Transformation: Building a Strong Data Foundation with AWS IoT SiteWise

Over the last few years, the industrial and manufacturing sectors have witnessed an accelerated transformation fueled by the advancement of the Industrial Internet of Things (IIoT), artificial intelligence (AI), and machine learning (ML). At the heart of this transformation is data, which when harnessed effectively, can propel businesses to new heights of operational efficiency, innovation, and customer satisfaction. Building a robust industrial data foundation is not just a strategic move; it’s an imperative for any manufacturer or industrial enterprise aiming to thrive in the digital era.

AWS IoT SiteWise is a managed service that makes it easy to collect, organize, and analyze data from industrial equipment at scale, helping customers make better, data-driven decisions. Our customers such as Volkswagen Group, Coca-Cola İçecek, and Yara International have used AWS IoT SiteWise to build industrial data platforms that allow them to contextualize and analyze Operational Technology (OT) data generated across their plants, creating a global view of their operations and businesses. In addition, our AWS Partners such as Embassy of Things (EOT), Tata Consulting Services (TCS) Edge2Web, TensorIoT, and Radix Engineering have made AWS IoT SiteWise the foundation for purpose-built applications that enable use cases such as predictive maintenance and asset performance monitoring. Through these engagements with customers and partners, we have learned that the main obstacles in scaling digital transformation initiatives include project complexity, infrastructure costs, and time to value.

To address these obstacles, we have recently launched new features in AWS IoT SiteWise that simplify how customers and partners apply analytics and AI/ML to industrial equipment data stored in AWS IoT SiteWise. The new features provide an up to 70% reduction in the cost to ingest data into the cloud, reduce project timelines from months to weeks, and make data more easily accessible for Business Intelligence (BI) dashboards and ML applications. These enhancements help customers onboard asset models and hierarchies faster, run analytical workflows within minutes of ingestion, and deploy predictive maintenance use cases faster to avoid unplanned downtime. With this launch, AWS makes it easier and more cost effective to transform large amounts of diverse industrial data into actionable insights, drive operational efficiencies, and improve decision making.

In this blog post, we dive into the details of the recently released features in AWS IoT SiteWise, as well as how AWS customers and partners are using these capabilities to facilitate the modernization of their data infrastructure.

Accelerating the Pace of Transformation

Standardizing visibility across operations is a key component of industrial transformation. It represents a move away from traditional, disjointed, and manual monitoring methods and requires an integrated, data-driven approach built on a unified view of contextualized data. AWS IoT SiteWise delivers this data standardization and context with asset models.  Models help organize the data and allow analysis at the enterprise, site, area, and machine level. However, given the complexity of industrial operations, building and maintaining models that accurately represent physical assets can be time consuming and delay time to insight.

With newly added APIs, AWS IoT SiteWise now allows you to bulk import, export, and update industrial asset model metadata at scale from diverse systems such as data historians, other AWS accounts, or – in the case of AWS Independent Software Vendors (ISV) Partners – their own industrial data modeling tools.

Import equipment metadata from external systems such as historians

Figure 1: Import equipment metadata from external systems such as historians.

In addition, AWS IoT SiteWise now supports the creation of asset model components and sub-components that customers can reuse to create new asset models. Asset model components let customers split complex machines into parts that are reusable across their enterprise. Customers can create a company-wide component library, driving model standardization and supporting more efficient scaling as their operations grow and become more complex. The figure below shows how a complex welding robot machine can be modeled using a reusable servo motor component. The new features shorten the time to onboard new industrial use cases from months to weeks, and accelerate time to value by ingesting data from various industrial data sources into a consolidated view faster.

Create reusable component models to describe your assets and organize data

Figure 2: Create reusable component models to describe your assets and organize data.

Creating a unified view of real time and historical equipment data

AWS IoT SiteWise provides secure, centralized storage for both real-time and historical equipment data. End users and industrial applications can consume data stored in AWS IoT SiteWise to gain valuable insights and drive business outcomes.

To collect real-time data from equipment, AWS IoT SiteWise provides AWS IoT SiteWise Edge, software created by AWS and deployed on premises to make it easy to collect, organize, process, and monitor equipment at the edge. With SiteWise Edge, customers can securely connect to and read data from equipment using industrial protocols and standards such as OPC-UA. In collaboration with AWS Partner Domatica, we recently added support for an additional 10 industrial protocols including MQTT, Modbus, and SIMATIC S7, diversifying the type of data that can be ingested into AWS IoT SiteWise from equipment, machines, and legacy systems for processing at the edge or enriching your industrial data lake. By ingesting data to the cloud with sub-second latency, customers can use AWS IoT SiteWise to monitor hundreds of thousands of high-value assets across their industrial operations in near real time.

To connect to equipment using supported protocols via integration with AWS Partner Domatica, configure your devices using their EasyEdge software

Figure 3: To connect to equipment using supported protocols via integration with AWS Partner Domatica, configure your devices using their EasyEdge software.

Not all equipment data is needed in the cloud in near-real-time, however. As we worked with customers in the energy, discrete manufacturing, and process industries, we learned that only 10% to 30% of equipment data sent to the cloud is used in near-real-time cloud-based dashboards.  The rest, 70% to 90%, is used in analytical applications, like BI dashboards or machine learning model training that only require data in the cloud within minutes, not seconds.  This provides us an opportunity to optimize in the way data is ingested and stored.

We recently announced the launch of buffered data ingestion to deliver the best cost and performance for data needed to support analytical use cases. With buffered ingestion customers can configure which data streams will be buffered at the edge before they are ingested to the cloud. This allows customers to reduce their cost of ingesting data to the cloud by up to 70%.

Cost efficient and optimized storage for analytical queries

AWS IoT SiteWise has multiple storage tiers that provide flexibility to support different use cases while balancing performance and cost efficiency. The hot storage tier is optimized for frequently accessed data, with low write-to-read latency for real-time applications such as interactive dashboards. The cold storage tier uses an Amazon S3 bucket to store data that is rarely used. Recently, we’ve also added a new warm storage tier designed for cost-efficient storage of historical data. It is optimized for retrieving large volumes of data with medium write-to-read latency for applications such as BI, reporting tools, and ML model training. This warm storage tier allows customers to retain large amounts of historical data at near Amazon S3 cost per GB storage prices.

Customers using the warm storage tier can also use the new Query API. The Query API lets customers retrieve metadata and time-series data from asset models, assets, measurements, metrics, transforms, and aggregates using SQL-like query statements in a single API request. This capability is compatible with tools such as Amazon QuickSight, PowerBI, and Microsoft Excel to power near real-time and historical enterprise performance reports.

Customers can explore their data and extract insights using SQL query statements with the new Query API. The following example shows how a user can query RPM information from all machines with “Engine” in their name.

select a.event_timestamp,b.asset_name ,c.property_name , a.quality,a.integer_value
from raw_time_series a,asset b , asset_property c
where a.event_timestamp > 1698335614
and b.asset_name LIKE ‘Engine%’
and c.property_name = ‘RPM’

event_timestamp asset_name property_name quality integer_value
26-10-2023T15:53:34 Engine001 RPM GOOD 2857
26-10-2023T15:53:34 Engine002 RPM GOOD 2549
26-10-2023T15:63:34 Engine001 RPM GOOD 2753
26-10-2023T15:63:34 Engine002 RPM GOOD 2349

Table 1: Retrieve data through queries using SQL statements.

Use machine learning to drive predictive maintenance programs

Recently, we have seen multiple customers merging their industrial equipment data from AWS IoT SiteWise with Amazon Lookout for Equipment to create machine learning models that can provide predictions and detect abnormal equipment behavior. This was a multi-step, somewhat time-consuming process customers had to go through. With the new native integration between AWS IoT SiteWise and Amazon Lookout for Equipment, we’re making it possible for you to directly sync data between these two services without building a complex set of integrations or writing any code. This allows you to easily build Lookout for Equipment machine learning models directly through AWS IoT SiteWise and go from reactive to proactive with anomaly detection and predictive maintenance.

For example, Toyota Motors North America (TMNA) has deployed models created in Amazon Lookout for Equipment using AWS IoT SiteWise data to their CNC machines.  With more than 200 CNC machines per site running 24/7, predictive maintenance was time consuming and costly for the TMNA Maintenance Team. TMNA has used AWS IoT SiteWise to develop a Predictive Maintenance solution capable of predicting failures days in advance, reducing unplanned downtime. Since deployment, the customer has been able to prevent dozens of accidents and hours of downtime, as well as improving operational availability by 10% vs. the previous 12-month average.

“The Operation Availability of our focus line was between 78-82%, incurring around 40 hours of downtime each month. With the help of AWS, we have found many problems in our machines, if left unnoticed would lead to critical failure. Now our OA is 92% and the downtime is around 20 hours!” – Braden Burford, Sr. Maintenance Engineer, Toyota

Contextualize equipment data to gain more powerful insights

Industrial transformation is largely centered around unlocking the potential of data from equipment, machines, and legacy systems. Traditional data management systems are no longer sufficient to meet the increasing demands for efficiency, scalability, and innovation. With these enhancements, AWS IoT SiteWise continues to deliver on its promise to provide a modern industrial data infrastructure that enables a scalable, unified, and integrated approach to harness data as an asset. It provides a cost-efficient, secure, and repeatable framework to make industrial datasets accessible to help customers build a strong foundation for industrial transformation and optimize their operations.

AWS customer Bristol Myers Squibb (BMS), a global leader in biopharmaceuticals, serves as a sterling example of how modernizing your industrial data infrastructure with AWS IoT SiteWise can transform your operations. With an ambitious goal to enhance business strategies across its Biologics, Pharma, and Cell-Therapy units, BMS recognized the need for an overhaul of its legacy data systems. Their primary objectives were clear: 1/ Achieve enterprise-wide visibility. 2/ Establish end-to-end traceability. 3/ Implement a single, validated enterprise solution for process monitoring, predictive asset maintenance, and continued process verification (CPV).

BMS turned to AWS IoT SiteWise for a consolidated approach to data management that would allow them to enhance visibility and analytics across their enterprise. By unlocking data from their Enterprise PI Historian and channeling it into a unified data lake on AWS, BMS achieved unprecedented scale, performance, and speed in data management.

One of the critical advancements for BMS was the ability to add context to their data by aggregating it with information from their Enterprise Resource Planning (ERP) and other systems. This provided richer site analytics for product batches being manufactured across various locations.

“In our quest for improved business strategies in Biologics, Pharma, and Cell-Therapy, enhancing visibility and traceability was crucial. AWS IoT SiteWise proved to be the perfect solution. By modernizing our data infrastructure with AWS, we seamlessly consolidated various data sources into a unified data hub, optimizing efficiency and scalability. This transformation allowed us to combine data from diverse systems and enabled insightful analytics for product batches across multiple sites. It significantly bolstered our ability to predict asset maintenance and shed light on newer potential use-cases. It’s a game-changer.” – Nitin Bhatti, GPS IT, Manufacturing Analytics at Bristol Myers Squibb

The transformation at BMS has set the stage for future innovations. With their modernized infrastructure, they are now positioned to explore additional use cases such as Predictive Asset Maintenance (PAM) and multi-variate analysis. The long-term vision includes extending the use and analysis of data beyond site personnel, providing a comprehensive, enterprise-wide view.

Delivering Business Outcomes in Collaboration with AWS Partners

Industrial companies going through digital transformation have found that scaling their projects is challenging. Taking initiatives from proof of concept to large scale enterprise deployments is resource intensive and demands specialized skills. AWS Partners have deep expertise across the industrial verticals and understand the drivers needed to generate long term customer value by offering solutions that solve line of business use cases. These partners help customers build a robust data foundation using AWS IoT SiteWise, and then use that data foundation to help customers solve their specialized use cases. A few examples of AWS IoT SiteWise partners are highlighted below.

EOT has built Twin Fusion, a suite of Software-as-a-Service (SaaS) products that use AWS IoT SiteWise to unlock, manage, visualize, and action their legacy IoT data with advanced analytics, ML, and Generative AI in the AWS cloud. Twin Fusion is part of the AWS Guidance for Industrial Data Fabric (IDF). Twin Fusion provides an end-to-end solution to ingest IIoT data and semantic data from machines and data historians into AWS IoT SiteWise. Twin Fusion provides an enterprise-wide digital twin graph asset model that fuses metadata from multiple industrial data sources. The product provides operational dashboards for end-user data analysis, asset hierarchy search, embedded ML model results, and enterprise-wide optimization of industrial assets using AI.

TCS are experts in modernizing historians with AWS services and they accelerate their customer’s time to value with AWS IoT SiteWise deployed at the edge and in the AWS cloud. TCS helps customers bring data from multiple historians into a single enterprise cloud historian, breaking down data silo’s to solve industrial challenges including optimized equipment downtime, improved cycle times, consistent production, defect reduction, and environmental compliance.

Edge2Web is using AWS IoT SiteWise as the foundation of its open platform suite of no-code and low-code industrial applications. Edge2Web applications help customers better manage asset fleets, reduce machine downtime, improve product quality, and optimize production performance.

TensorIoT has created the SmartInsights solution built on AWS IoT SiteWise. SmartInsights provides robust visualizations of ‘what has happened’ and ‘what is going to happen’ in a single pane of glass. SmartInsights enables customers to solve use cases such as predictive maintenance, remote asset monitoring, and renewable asset performance prediction and maintenance.

Radix Engineering is focused on helping industrial customers unlock timeseries data stored at the edge and modernize their legacy industrial operational technology (OT) architecture with AWS IoT SiteWise while driving improved operations and reliability with integrated machine learning (ML) models and insights.

Each of these partner solutions not only addresses specific industrial challenges but also showcases the vital role of specialized expertise and advanced tools such as AWS IoT SiteWise in successfully scaling digital transformation initiatives for long-term business value and efficiency.

A Blueprint for Transformation

The success stories from Toyota Motors North America and Bristol Myers Squibb serve as a blueprint for other enterprises. These leaders and many more have embraced AWS IoT SiteWise as the service that provides a scalable and repeatable industrial data foundation, integrating it into their daily operations and are harnessing the power of historical and real-time equipment data to realize the value of digital transformation.

Click here to get started with AWS IoT SiteWise and, if you’re attending re:Invent 2023, make sure to join the below sessions to dive deep into these new capabilities.

IOT206 | Accelerating industrial transformation with IoT on AWS

IOT215 | Accelerate shop floor digitization with edge-to-cloud data integration

IOT212 | Modernizing your data historian with AWS IoT SiteWise

IOT203 | Automated anomaly detection for smart manufacturing