How To Use IoT Data To Build a Smart Factory

As the manufacturing industry continues to evolve, businesses are exploring new ways to use technology to increase efficiency, productivity, and profitability. One way is through the use of Internet of Things (IoT) technology. By collecting data from various sensors and devices, IoT enables manufacturers to gain real-time insights into their operations and make data-driven decisions.

However, the sheer amount of data generated by IoT can be overwhelming for SMB leaders and IT managers who are driving the smart factory initiative. Without the proper tools and strategies to analyze and visualize this data, it can be challenging to find the type of insights you really need to make better decisions about the future of your business.

In this article, we'll explore the steps and best practices for using IoT data to make your factory a smart one. From understanding the types of data generated by IoT devices to analyzing and visualizing this data, we'll provide practical tips that can help SMB leaders and IT managers enable the creation of smarter, more efficient factories.

Let's get started.

What are the current challenges in using IoT data?

Research shows that while IoT data presents tremendous opportunities for manufacturers, it also comes with a set of challenges. [1] One of the biggest challenges is the velocity and volume of data generated by the vast number of sensors and devices deployed in a smart factory. Millions of IoT endpoint devices are added daily, creating a network effect that exponentially increases the value and volume of data and analytics.

The skills gap in data analytics is another significant challenge for SMB leaders and IT managers. They need the technical skills and tools to handle and analyze this data in order to extract truly valuable insights. Without them, it can be impossible to get what they need from the massive amount of IoT data generated daily.

Another challenge is the absence of an IoT-enabled data lake. Without a data lake in place where both IoT and manufacturing data can be collected, contextualized, and visualized, all modernization projects for improving processes and automation will be for naught. Not properly curating the IoT data with other manufacturing data will result in incomplete and inaccurate insights, leading to increases in operational expenses, security vulnerabilities, and poor data quality.

Steps for using IoT data for enabling the creation of a smart factory

With the rise of IoT investments in manufacturing plants, a significant amount of data streams are now being collected. All IoT data is telemetry, meaning that it is produced by sensors and other endpoints, and the data has a low level of contextualization, making it difficult to derive insights from the raw data alone. The data will also most likely be raw, unfiltered, and repetitive.

When IoT data is first collected, it is referred to as "hot" data. However, as it gets processed over time, an increased level of contextualization is added, primarily by blending manufacturing data with the raw IoT data, which helps transform raw IoT data into warm or cold data. By adding contextualization, IoT data can be used to discover insights that can help a business make better decisions and can be visualized using data visualization software. Here are the steps in more detail:

1. Acquire: Implement a data collection engine

This first step is collecting the data from the various data streams, including time series events, messages, transactions, and often repetitive data that comes from a variety of IoT sensors. Raw endpoint data flowing through an IoT architecture is typically high in volume, velocity, and variety. At the aggregation point, IoT data is tagged as "hot."

To implement this step, you will need a data collection engine. The data collection engine is responsible for ingesting data from various sources and integrating it into a central data repository. The data collection engine can be a software or hardware solution that gathers the data from your factory's various IoT devices and sensors. This data is in its raw form and needs to be processed to gain useful insights.

Data collection software should be able to handle the high volume of data and be scalable to accommodate the increasing amount of data that will be generated as the smart factory grows. It should also have the ability to integrate with different types of IoT devices and sensors that your factory uses to ensure that all data generated can be captured and processed. Another feature your data collection engine should have is the ability to handle real-time data processing and provide alerts in case of anomalies.

Once the data is collected, it needs to be stored in a central repository. This repository is where the data will be stored for further processing and analysis, which we will look at in the next step.

2. Contextualize: Establish a data lake

Once the raw endpoint data has been acquired, the next step is to contextualize it. This is where common data lake functions come into play, such as aggregation, tagging, alerts, controls, classification, clustering, detections, rule mining, and filtering.

Contextualization is a critical step that involves blending manufacturing data with IoT data. This allows the data to be tagged as either warm or cold, depending on the level of sophistication needed for analysis. As the data goes through this contextualization funnel, an increased level of contextualization is added, making it easier to analyze and extract value from the data.

Metadata management software plays a crucial role in this step. The software will create the tags, labels, and classifications that make it easier to find, use, and manage the data. By creating metadata, the data is more easily searchable and accessible to those who need it. This software also ensures the accuracy and consistency of metadata across the data lake. With proper metadata management, data can be easily tracked and managed throughout its lifecycle.

Cloud storage software is another important part of a data lake. It will give you a secure, centralized repository for storing and processing data which can be accessed from anywhere, at any time. Cloud storage software also makes processing large amounts of data quick and efficient, which is essential to a smart factory because real-time data processing can help business leaders optimize the manufacturing process quicker.

3. Visualize: Generate business insights

In this step, the contextualized data is turned into descriptive/diagnostic, predictive, or prescriptive insights that business leaders can use. The best way to provide these insights to non-technical leaders is by providing them a way to visualize the insights, so they can quickly understand the complex data.

For this step, you need data visualization software. Data visualization software provides a graphical representation of the IoT and manufacturing data you've collected and contextualized, making it easier to find patterns and trends, keep an eye on important KPIs, and track progress toward business goals. This software allows businesses to analyze large amounts of data quickly and efficiently, and it enables decision-makers to identify actionable insights that can improve operational efficiencies and identify new revenue opportunities.

With data visualization software, you can create interactive dashboards, charts, and graphs you can customize and share with other team members. These visualizations will help communicate complex data in an easy-to-understand format, making it simpler for leaders to find trends and make decisions.

Defining hot, warm, and cold data

In general, manufacturing data can be categorized as hot, warm, or cold based on its relevance and utility for decision-making purposes.

  • Hot data: Hot data is the rawest form of data, which is consumed by IoT analytics software for condition-based maintenance (CBM) applications. It has a low level of contextualization and is typically used to monitor a manufacturing process in real-time. An example of hot data would be monitoring a production drill's temperature to avoid overheating, where the drill bit's fault or foreign material can be investigated by functional leaders.

  • Warm data: Warm data is historical data obtained from a database, manufacturing execution system (MES), or statistical Process control (SPC) system. It is typically used for predictive maintenance applications and has a semi-contextualized level of data. For example, a supply chain leader might use warm data to determine a plant's production yield over time to set realistic delivery expectations for the end customer.

  • Cold data: Cold data is batch data that is blended with archived, business intelligence, and operations data. It has a high level of complexity and is typically used for making company-wide decisions by C-level leaders. Cold data is relevant to decision-makers who need to understand the overall health of a manufacturer's plants in different areas. Cold data requires high-level contextualization. Its insights can be derived by blending batch IoT data with fragmented operational systems' data, such as financial, quality, or resource planning data.

Recommendations for implementation

Leveraging IoT data in smart factories is crucial for businesses to remain competitive in the ever-changing manufacturing industry. Here are some recommendations for SMB leaders and IT managers to follow when implementing IoT-enabled data lakes in their smart factories:

  • Develop a digital transformation alignment plan to determine which manufacturing asset to start integrating data.

  • Develop a data schema framework prior to building an IoT-enabled data lake.

  • Categorize the hot data into its proper category and tag the contextualized IoT data to either warm or cold as per the data schema requirements.

  • Identify common use cases required for each key decision maker and map the type of IoT data needed for each use case.

  • Collaborate with each department leader to build the rule-based reporting library of metrics that each decision-maker requires.

  • Map out the types of KPIs and data insights that each critical decision-maker requires.

  • Audit the types of key metrics, reporting, and presentations each departmental decision-maker is consuming and then develop an action plan to end-of-life any conflicting legacy data presentations and/or KPIs.

To learn more about IoT devices and how to use the insights they provide to make more well-informed decisions, visit these resources: