Data and Analytics in Manufacturing: Overcoming Challenges

Introduction

Manufacturing firms globally are grappling with constantly evolving consumer behavior, their preferences, and a hyper competitive market scenario.

International geo-political factors and competition from cheaper sources have further complicated the market and intensified the need for manufacturers to integrate, automate, and deploy smart technology on their shop floors, embracing Industry 4.0 best practices. Improved productivity and operational efficiency, minimal equipment downtime through machine learning (ML)-led preventive and predictive maintenance, resilient supply chains through improved forecasting and digital auditing, and more, are some obvious benefits of digitalization. Automation of routine plant production and maintenance with minimal human intervention activities also free up manpower hours and executive time to focus on innovation, research and development (R&D), and strategy.

Digital tools for manufacturing transformation

Capabilities such as remote plant operations management via internet of things (IoT) and 5G integration with smart technologies, including machine learning (ML), digital twin, and mixed reality in manufacturing, have become a necessity not just for business continuity but even for innovation, personnel training, and informed executive decision-making.

Data and analytics will play a critical role in achieving this transformation, but manufacturers face numerous challenges.

Data and analytics constraints impacting manufacturers

While most manufacturers realize the imminent need for a robust data and analytics strategy, they face challenges adopting and implementing them.

We have summarized the challenges into five buckets, as described below:

• Embarking on the data and analytics journey: Currently, the level of ambiguity while dealing with data and analytics is extremely high for many manufacturers. Within the manufacturing ecosystem, plant managers often put forth many hypotheses to increase productivity but are typically unsure of what might actually work. Analysts and data scientists highlight the complexities in data quality which require hundreds of manhours involving manual cleaning and sorting of data. IT stakeholders are focused on technological challenges and limitations that need to be considered while finance is concerned with cost impact. So, does a firm need to address all issues before gaining value out of the investment made? The answer is obviously no. However, manufacturers need to have a strategy in place with a middle path approach, to address all concerns and ensure key stakeholder endorsement on core issues to be addressed as they start off on their transformation journeys.

• Managing data: Supervising data management and governance models across different business processes and system complexities in an organization is a key issue. Plant analytics requires a variety of data from various sources:

a. Structured data from enterprise resource planning (ERP), manufacturing execution systems (MES), product lifecycle management (PLM), supervisory control and data acquisition (SCADA), and others.

b. Time series data from data historians.

c. Unstructured data from various social media and market research reports.

The challenge of managing diverse data from disparate sources is that integrating this data simultaneously can make the process too complex. Also, rigid governance processes and fixing quality issues, which are time consuming, can diminish the value of data if not delivered at the right time

• Technology architecture for plant data lakes: Given the complex legacy systems that exist in most plants, defining a holistic and contextual architecture using the most relevant technologies is another area of struggle. Even with cloud technology, costs and the total cost of ownership (TCO) need to be carefully evaluated. Ensuring flexibility in the architecture that allows for on-demand scale-up and downtime is key while defining the architecture.

• Project execution from a people and process perspective: Many industry experts recommend executing projects in an agile manner. But in a plant, implementing projects using agile has to be thought through in detail to anticipate various complexities. Many agile projects in plant data and analytics programs have failed because of data and other complexities that are not typical in projects executed for finance or sales.

• Defining and measuring success: Establishing a value measurement framework upfront and defining expectations and key result areas (KRAs) are critical. Despite executing seemingly successful projects, many manufacturers perceive that investments made in data and analytics programs do not result in the intended return.