Data Flow Architecture

Vision

Data volumes in electronics manufacturing facilities are growing at exponential rates. Pervasive data velocities measured as the data collection rates from equipment increased from less than 1 Hz in the 1990s to 10 Hz in 2020 and are expected to be at 100 Hz in a few years with some dedicated sensor data (e.g., IoT) collection rates in excess of 10 KHz. Delivering higher data volumes at real-time and near-real-time rates will increase the availability of equipment parametric data to positively impact yield and quality.

Data from equipment, maintenance, yield, inventory management, manufacturing execution system (MES) and enterprise resource planning (ERP) existed for several years; however, analytics tools are evolving to leverage and merge multiple data sources to investigate relationships, detect anomalies, and predict events.

Scope

The emergence of big data in electronics manufacturing operations is discussed in terms of the “5Vs Framework”, as follows:

1. Volume (how much data in absolute terms),

2. Velocity (how fast is that data being transmitted between elements),

3. Variety (or data merging from multiple, often heterogeneous, sources),

4. Veracity (or data quality), and

5. Value (or application of analytics, automation and control).

The “5 Vs” are foundational to appreciate widespread adoption of big data analytics by the electronics manufacturing ecosystem.

Technical Needs, Gaps and Solutions

The technology issues surrounding data flow architectures, the associated needs, technology status of those needs, as well as gaps and challenges to overcome, are summarized below. The time period considered is from 2023 to 2033.

Technology Status Legend

For each need, the status of today’s technology is indicated by label and color as follows:

Definitions for “Gap,” “Challenge,” and “Current Technology Status” are below:

Table 1. Smart Manufacturing Data Architecture Needs, Gaps, and Today’s Technology Status with Respect to Current and Future Needs

Approaches to address Needs, Gaps and Challenges

Table 2 considers approaches to address the above needs, gaps and challenges. The evolution of these is projected out over a 10-year timeframe using technology readiness levels (TRLs).

Table 2. Data Flow Architecture Potential Solutions

In electronics manufacturing facilities, an intelligent, flexible factory floor requires flexible, resilient, low-latency communications between

(a) AI-enhanced factory floor control systems and enterprise management systems and

(b) manufacturing/process equipment, material flows, and building facilities.

The trend to move computing closer to the edge will remove some latency. Communication connectivity for equipment, material flows, and facilities will also need the low latency and high reliability of wired Ethernet, but delivered wirelessly. 5G solutions are emerging to address this, but the cost point and power consumption remain high. Other solutions, such as industrial 802.15.4¹-based protocols are much more power efficient and cost effective, but lack the required performance, particularly in terms of resiliency, because of their use of shared spectrum.

The value proposition for a comprehensive data flow architecture supporting a wide variety of manufacturing facility data types is well understood, but there are a number of barriers that industry needs to address for such solutions to become ubiquitous:

• Data collection rapidly outpaced data integration, architecture, and analytics. Today, vast amounts of data are going unused.  Data is only helpful if it is leveraged for actionable insights which lead to more efficient processes and ultimately positive ROI.

• A range of research and industry experience demonstrate that between 60 and 90% of all Industry 4.0 pilot or Proof of Concept (PoC) projects never progress beyond the initial stage. Often driven by technical rather than operational teams, little focus is given to how to scale-up if the pilot is a success.

• Most companies fail to form an early view on the comprehensive target-state technology stack that is scalable and enabled for analytics.

• Industry needs to address the lack of harmonization for factory-floor data formats, both for machine-to-controller and machine-to-machine.

• Smaller companies will struggle to adopt best practices and cutting-edge solutions unless installation, configuration and management are greatly simplified. Architectural simplifications, e.g. embedded analytics, will help.

• It will be a return-on-investment challenge to incorporate legacy equipment into the overall data flow architecture.

In manufacturing, execution is everything. Manufacturing leaders must shift their target from mere data collection to data-informed execution. To do so, this begins with creating data integration at every level of operations, whether between shop floor machines or top floor enterprise systems. This integration must be done with proper data governance and architectures, centralizing data into data lakes which ultimately allow for high quality analytics across multiple departments. Applications, data management, infrastructure needs, security and edge process/control must be considered to select the right platform.

1. IEEE Standards Association, “IEEE Standard for Low-Rate Wireless Networks“, https://standards.ieee.org/ieee/802.15.4/7029/, 2020/

Return to topic overview

Go to Data Architecture Vision