Did you know that data analytics can help manufacturers address most shopfloor problems proactively (remotely), and enhance their delivery schedules?

Target audience: Decision makers (CXOs/Directors) of manufacturing & allied firms, management consultants, business strategists, innovators, and curious people.

Reading time: 5-10 min.

Introduction:

Manufacturing analytics is the use of data collected from operations and incidents to improve quality, enhance worker performance, increase throughput, reduce overall costs, and optimize supply chains.

Generally, required data is collected via sensors and Industrial Internet of Things (IIoTs) linked to machines and industrial robots, and operator inputs like computers or Human Machine Interfaces (HMIs). This data is centrally stored on local hard drives or cloud servers for analysis.

Product Quality:

Production efficiency depends on the quality of each batch manufactured in a factory. Some important tools widely used to monitor the quality of each batch are Statistical Process Control (SPC), and Six Sigma.

Control charts are one of the most important SPC tools among the seven quality control (7-QC) tools. These charts can be created with sample data from multiple batches to monitor process changes over time. An investigation into the production processes can be initiated when any of the “Out-of-Control Signals” are flagged.

Generally, “Six Sigma Quality” is defined as a production process which is controlled in between ±3s (3 sigma) from the centre line in a control chart, though the tolerance limits may be set to ±6s. In mathematical terms, Six Sigma can be defined as 3.4 defects per million parts produced. One of the widely used Six Sigma methodologies to solve problems is DMAIC: define, measure, analyse, improve, and control.

Use of automated data collection systems across the factory would help a plant’s quality manager respond to an ‘out-of-control’ production process in real-time. A Supervisory Control and Data Acquisition (SCADA) system can be used to collect data and assist manufacturing analytics systems. In most cases, reliability of this data (automated collection) is higher than physically collected data.

Is there a need for Data Analytics?

In the past decade (2010—2020), there has been an emphasis on regional growth, nationalism, and reduced dependence on a single supplier/country/region for products and services. The COVID-19 pandemic has increased the level of urgency towards smart manufacturing systems. Since late 2020, there has been a trend of high demand (consumer products) and rise in multiple supply shocks. These include—chip crunch, container shortage, truck driver and other labour shortage, natural calamities (Texas freeze, flash floods), and electricity shortage.

Reasons to incorporate Technology in Manufacturing:

• Availability of customised industrial automation systems for SMEs which are moderately priced (electronics, software systems, programmers).
• Challenges of global supply chains (delays, surge prices, acute shortages) over regional supply chains.
• Rise of stringent regulatory compliances—better worker conditions, component traceability,  and product quality.

Current & Future Trends:

As per Markets and Markets, the global product analytics market size is poised to grow exponentially. This market is expected to grow from USD 9.6 billion in 2021 to about USD 25.3 billion in 2026, at a CAGR of about 21.3%. North America would have the largest market share over this period, led by Canada. Manufacturers are the leading end users (2021), while the top three sectors include—Retail and Consumer Goods, and Manufacturing.

In today’s highly competitive manufacturing era, a plant manager’s response time is shrunk down significantly. For instance, failures in a production batch may require a thorough investigation within a few hours to ensure orders are not delayed. Use of Lean manufacturing and Just-In-Time manufacturing (JIT) systems require a high level of synchrony along the entire manufacturing ecosystem (value chains). This is in part due to low inventory levels at each manufacturer in these chains.

Productivity improvements through manufacturing analytics:

• Real-time Inventory Management—
  • Manufacturing plants
  • Suppliers
  • Transported parts (IIoTs, sensors, 4G and 5G networks)
  • Reduced operational costs (production delays)
  • Enhanced delivery schedules (accurate forecasting)
• Preventive and Predictive Maintenance of machines across production floors (IIoTs, sensors) to improve their lifespans, and to improve production efficiencies (reduce breakdowns and bottlenecks).
• Increased Visibility of production timelines across value chains (networked ERP systems), and to accurately schedule raw materials.
• Improved Demand Forecasting to ensure production goals are achieved and customers receive their products on time.

Potential Risks and Threats:

Critical challenges to rapidly deploy analytical tools in manufacturing:

• Data and Scope Issues—Costs involved in reliable data collection (HMIs, sensors, IoTs) and storage (HDDs, servers, cloud services) have reduced significantly in the past few years. In many cases, these changes along with a lack of clarity (business advisors) have led to collection of large volumes of unusable and poor-quality data. It is important to appropriately understand business problems and choose incremental projects for a high chance of success. Coordination of multiple stakeholders for project selection (scope) is also essential for its successful implementation.
• Skills Shortage—Today, there is an acute shortage of reliable data scientists. Many young graduates can create a pool of models from a data set, which may not provide useful insights. A skilled data scientist should have a combination of required technical skills, business expertise, and domain knowledge. Ideally, such workers would require years of industry experience and management expertise, along with formal education in data science, and engineering.
• Actionable Results—A well-known adage “Garbage In Garbage Out” (GIGO) is applicable to data analytics too. Choosing a wrong project or collecting unreliable data can result in huge losses (time and money). To avoid such situations, project leaders should choose actionable and agile projects. These projects should be reviewed periodically to ensure data collection is accurate, and insights obtained are aligned to business priorities.

Conclusion:

To sum up, data analytics is a double-edged sword which requires a good strategy (long-term vision), and an agile approach (incremental projects).

Often, business leaders choose a large project over multiple small ones—it would require less hands-on monitoring and be supported by multiple department heads. Unfortunately, disruptive changes in present markets could make new technologies (hardware and applications) obsolete in a few years after their launch.

For example, today’s advanced  artificial intelligence and machine learning systems can provide game-changing solutions to business problems in a matter of months. Therefore, it would be wise to use multiple small projects (sprints) to address today’s issues, and be prepared (proactive approach) to face new manufacturing challenges.