For decades, EMS factories have collected and used machine data to unlock process improvements. This blog will illustrate three common ways that tapping collected machine data helps improve operational efficiency and how this data empowers trail blazing advanced analytics for Industry 4.0.
Modern EMS factories are filled with machines that produce rich data about operations. The richness, format and content of this data varies by machine vendor, type, and even model. Generally, the offered data includes detailed information on operations performed, faults encountered during processing, materials consumed, products produced, and active recipes.
Three Common Values from Data Collection for Manufacturers Today
Machine data creates value in these three ways:
- Process monitoring with machine data allows fewer operators to monitor more machines for issues — resulting in fewer errors at a lower cost.
- Process interlocking eliminates common operational mistakes — using machine data and application of business rules to ensure proper setup.
- Performance tracking watches what percentage of a facility’s installed production capacity is being used — helping to avoid costly downtime, reduce pressure on production bottlenecks, and guide asset management decisions.
All three of these approaches to the data are proven to improve operational efficiency (OE). To capitalize on these gains, more and more manufacturers seek forward-looking ways to leverage their machine data through Industry 4.0 initiatives, and technology is rising to that challenge.
In addition to examining these three ways machine data is improving OE, we will illustrate how advanced machine data analysis is changing the game—providing the ability to infer where problems may lie in the factory and which of those problems would be most impactful to solve.
Process Monitoring (Real-time Operational Monitoring)
Machine data speeds production and increases efficiency for individual processes on the factory floor. With the standard data provided by many machines, a small number of operators can monitor a fleet of machines.
Instead of putting the human machine interface (HMI) — a fancy way of saying monitor and keyboard — on the machine itself, remote operators can monitor a large number of the machines in the production facility from one place. Error resolution still takes place at the machine level, but it happens faster and with fewer people involved.
Real-time monitoring increases operational efficiency by reducing the time between a fault occurrence and resolution but does not decrease the total number of faults. More advanced versions of real-time monitoring don’t wait for problems before notifying operators about potential trouble. Alert dispatching systems and interactive dashboards analyze the performance of different factory lines and compare them against key performance indicators (KPIs), offering the chance to optimize performance instead of just using a break-fix model.
Process Interlocking
Process interlocking identifies efficiencies and lowers errors across lines and machines.
Factories are increasingly called upon to produce more diverse mixes of products, each with a high degree of complexity. Optimizing the interaction between machines and lines increases profitability and adds a competitive margin. Process interlocking uses machine data and applied business rules to make sure human setup mistakes are minimized. The intent is to increase the overall probability that everything is set up correctly to build a product.
To improve accuracy, process interlocking requires each machine to ask an external factory system to validate operator actions before the machine will accept them. For example, a machine could check a serial number printed on each circuit board (PCB) before placing components on that PCB to ensure that it belongs to the expected work order for that factory line.
Performance Tracking
Factories can improve OE over time by tracking key performance indicators (KPIs) based on machine data captured during factory operations. In the complex modern factory, well-designed numerical measures simplify understanding and decision making.
With globally collected machine data, overall equipment efficiency (OEE) statistics quantify what fraction of a factory’s installed production capacity is utilized. A perfect OEE score of 100% means the factory produces defect-free products as fast as physically possible. Historically, this has been a manual process that delivers KPIs often delayed by weeks or months. Automatically collecting and aggregating machine data allows the rapid calculation of stats like OEE, first-pass yield, and material attrition. These critical insights give business owners the ability to compare factories and see the effects of initiatives.
Since factories are large, complex operations, there is a desire to simplify this complexity down to a few numbers that allow humans to quickly understand and compare overall performance without needing to reason about large sets of complex parameters. Essentially, systems of reporting. However, advancements in both technology and expertise no longer necessitate simplified numbers for easy comparison and now allow for analysis far beyond collecting and reporting.
Advanced Analytics: Changing the Game Entirely
Advanced machine data analytics, hereafter just called advanced analytics, looks at the data very differently– because it can.
Powered by big-data techniques and supported by correct data architecture, advanced analytics looks at how machine data values change over time to infer where problems are in the factory and which problems are most impactful to solve.
Unlike real-time operational monitoring, the goal of advanced analytics is not to find and quickly resolve minute-by-minute issues that occur and disappear rapidly, but rather to identify larger structural problems that reduce efficiency day-after-day or month-after-month.
There is additional value in collecting and storing rich machine data for historical analysis because modern data analysis tools and methodologies have now made it possible to both efficiently collect and analyze this data.
Advanced analytics identify large structural issues and improve efficiency.
Asking the Data “Why”?
One of the most important aspects of this new category for data use for manufacturers collection and analysis, one where all the data is collected and stored in its native format for immediate, future, and historical use, is that it allows manufacturers to glean much more context from their data.
The majority of the tools available to manufacturers today, like the MES and other connectivity tools, only give machine operators enough data to understand what has already happened to their machines in the past. They are designed to be reporting tools to facilitate the current state of factory management.
Factories that rely solely on tools like MES are limited in a few critical ways. First, the type of data pulled from the machines is pre-defined for a specific purpose. It is designed to solve a particular set of problems and is not future-proofed against others. Because the context behind the reporting is not collected or structured, the data can’t be queried beyond what it’s already providing in its reports.
By contrast, this new approach to data, one where all the data is collected and stored, leaving nothing behind, provides a rich tapestry of context behind the basic reporting on KPIs. Once a machine has alerted on one of these KPIs such as attrition, downtime, or quality, manufacturers can now ask their data why something happened and can get an answer– without having to go to the floor of the factory to do it.
This approach provides value at every level of the organization and allows immediate access to it through intuitive analytics platforms.
The global view of performance tracking does not contain the information from the machine level of process monitoring; the behaviors of individual machines are aggregated to focus on the question of historical improvements. Leveraging modern data tools, manufacturers can perform advanced analytics that provide both a broad view and detailed data. With that view, large structural changes can be made to improve outcomes.
Real-time monitoring, performance tracking and process interlocking are proven methods for improving operational efficiency. With the right data, personnel tasked with machine, factory, and global efficiency can ask the data the right questions and take action on those answers to trigger dramatic operational improvement at every level.
To learn more about the strategic uses of machine data, read our technical paper.