Need continuing education credits? Join Us For Five Days of Education on the Industry's Leading Topics beginning October 5th!Save Your Seat
Virtualization, Cloud Analytics

Cloud-based analytics for manufacturing

Industrial organizations already are realizing the benefits of cloud computing for IT and business workloads, providing a path to manufacturing deployments.

By Megan Buntain March 11, 2020
Courtesy: Seeq

Cloud computing is recognized by industrial organizations as a key enabler for storing and analyzing data in volumes that seemed unfathomable even a few years ago. This is important because many organizations start and end their digital transformation strategy with the idea that better use of vast amounts of financial, customer, supply chain and operational data will improve operational efficiency and create new business models.

If cloud-based analytics tools are not in use today by a manufacturer’s process engineers, another department is likely using them. For example, an HR department is using software to analyze employee data, or analysts in sales and marketing are working to review demand for specific products and services. While these are considered productivity or business intelligence applications, they are also the entry point for cloud-based analytics in most companies.

To implement digital transformation strategies, there has been a rise of cloud-based analytics — applications, tools, and techniques deployed in the cloud instead of on premises — enabling organizations to quickly gain data insights. The cloud is streamlining this process because it enables rapid insights on much more data of different types with the near infinite scalability of computing resources. This empowers employees to solve an increasing array of complex business challenges in near real-time.

Defining cloud-based analytics terms

Cloud-based analytics is a broad term referring to several layers of computing capabilities. First is the underlying cloud infrastructure, the operating system and hardware layers, required to host data and applications in the cloud.

On top of that infrastructure is the data management layer, in a cloud service or a data lake, where various types of data are stored including but not limited to structured and unstructured text-based data, video data, and streaming IoT data. Applications in the analytics layer make use of this data. Calculations are performed to supply the visualization layer with the information required for trending, reporting, dashboards and other insights.

In manufacturing, analytics has traditionally been done on premises using a combination of historian data and spreadsheet analytics for ad hoc diagnostic, predictive, or operational dashboards for the plant, but that is changing as cloud-based analytics’ benefits are being realized.

The cloud can be used to make data available to the machine learning algorithms used in advanced analytics applications. Courtesy: Seeq

The cloud can be used to make data available to the machine learning algorithms used in advanced analytics applications. Courtesy: Seeq

Expected benefits

Applications deployed in the cloud benefit from core cloud capabilities. First is the cloud’s rental model versus the capital cost associated with hardware and infrastructure. A company’s information technology (IT) department no longer has to provision and maintain expensive servers to host these applications, and the result is a “pay as you go’” model where computing resources are spun up and down on demand.

One example is a web store that does 90% of its business in the weeks leading up to the holiday season. Before the cloud, that retailer had to purchase enough servers to handle the website traffic for the “burst” of demand at the peak, while remaining largely idle the rest of the year.

In an industrial context, as subject matter experts (SMEs) leverage new analytics tools to gain more insight into their operational data, more flexibility is required. SMEs may want to analyze new data sources such as operational data and contextual data, and organizations may need to make analytics tools available to more SMEs and other users to enable better collaboration and decision-making.

Another driving factor for cloud adoption is the ability to explore new types of analytics such as making historical and near real-time process data available for machine learning (Figure 1). Many manufacturers don’t want to run machine learning models within their real-time control systems, but they do want to take advantage of these and other advanced capabilities for improving product quality, predicting optimal maintenance windows to prevent unplanned downtime and other purposes. Copying operational data to the cloud makes it available for machine learning, enabling new analytics models to be explored without the risk of impacting the source production data or any existing applications relying on that data.

Cloud-based analytics makes it easy to break down data silos so users can access and connect to data regardless of its source. Once those silos are connected via the cloud, SMEs and other users (Figure 2) can scale up analytics to sites around the world and create ways of viewing global operational reporting to ensure the best possible business impact is achieved.

Employees in different departments can access cloud-based data and analytics worldwide. Courtesy: Seeq

Employees in different departments can access cloud-based data and analytics worldwide. Courtesy: Seeq

Cloud analytics: Getting started

When implementing cloud-based analytics, it’s important to begin with the end in mind. Too often energy and manufacturing companies spend considerable time planning for and migrating data and applications to the cloud, only to ask, “What now?” after the data is moved. Moving data or aggregating data in a cloud data lake doesn’t make it more valuable; it’s a step along the way to implementing a comprehensive data analytics strategy.

The surest way to avoid this outcome is ensuring SMEs are involved early in any analytics project. Only people with deep process expertise and a view of the unique impact of individual units within broader operational procedures can ensure this data leads to insight and productive action. The more SMES are provided with relevant, easy-to-use, flexible analytics applications, the more rapid return on investment.

One note of caution in operational data and cloud computing models is IT teams must resist the temptation to summarize process data in the cloud or apply business rules before connecting cloud analytics application to the data. When data is summarized, someone without direct knowledge of the asset or process is predetermining what SMEs might be interested in exploring, which can diminish its potential impact. The best practice here is store all the data in its native form so SMEs others can make decisions at analytics time about how and what to modify, for example data cleansing and access to any data set for investigation and model development.

SMEs use Seeq to directly interact with data of interest and find insights. Courtesy: Seeq

SMEs use Seeq to directly interact with data of interest and find insights. Courtesy: Seeq

Cloud success stories

An energy company with more than 50 operational sites dispersed across a large and geographically challenging environment was able to implement cloud-based analytics successfully by starting with six engineers at one site. These engineers identified three use cases: asset integrity monitoring and performance trending, predictive maintenance and production forecasting.

Within 90 days, the team had scaled up to more than 50 engineers in more than 10 locations working with these use cases to connect new sources of data by leveraging shared cloud deployment and collaboration capabilities. This unlocked the creativity of these engineers to find dozens of additional use cases to finding insights and improve asset availability and production outcomes (Figure 3).

By starting small and leveraging the cloud to scale up analytics quickly, the project owners measured the business impact, building their business case to get more users and sites on board.

The next step for the organization is to connect additional IoT data from their remote locations, keeping two critical factors in mind, network latency and analytics performance. Users won’t use cumbersome analytics tools with poor performance and excessive delays, so these factors must be addressed. Before cloud implementation, reporting from these remote locations was offline and manual.

To solve these issues, the company is using a hybrid cloud approach with edge analytics, which maintains some computing and analytics resources at the edge or close to the data source with results delivered to the cloud when the network is available.

The manufacturing industry is still in the early days of cloud-based analytics, but critical implementation learnings are emerging. Companies must keep SMEs at the center of any analytics effort, and they can leverage the cloud to scale up and down to connect data silos. Raw data must be kept whole to ensure analytics and insights are flexible. Collaboration across teams and sites also must be ensured to realize returns and broaden the potential business impact.

Coupling the right advanced analytics software with cloud platforms already in use by most organizations will help yield benefits in terms of improved operations.

Megan Buntain, director of cloud partnerships, Seeq Corp. Edited by Chris Vavra, associate editor, Control Engineering, CFE Media and Technology, cvavra@cfemedia.com.

MORE ANSWERS

Keywords: cloud-based analytics, cloud software

Cloud-based analytics allow manufacturers and subject matter experts (SMEs) to receive data faster.

SMEs can leverage the data and find unique insights to improve operational efficiency.

Cloud-based analytics makes it easy to break down data silos so users can access and connect to data regardless of its source.

Consider this

What benefits could your plant gain from cloud-based analytics?


Megan Buntain
Author Bio: Megan Buntain is the director of cloud partnerships at Seeq Corporation, a company building advanced analytics applications for engineers and analysts that accelerate insights into industrial process data. She was formerly a consultant with analytics, IoT, and blockchain software and services companies, and prior to that worked at Microsoft for 15 years.