What are the best practices for MES data analysis and visualization?

Close collaboration with business users is crucial for success here. They possess firsthand knowledge of the operational complexities and strategic goals driving the manufacturing processes. Moreover, their engagement will encourage a sense of ownership and therefore enhance adoption.

While defining goals and metrics is essential, there are other critical questions to consider: Who will access and visualize the data? Are people with different responsibilities looking at the same data? Can KPIs be tailored for each team? MES, by its nature, democratizes data access, bridging the gap from connectivity to visibility (ACATEH maturity index). It's vital to ensure that different teams have access to data that meets their unique needs. For example, site operations focus on daily production metrics, while the S&OP team might prioritize long-term demand and capacity planning.

First and foremost we need to identify the KPIs by engaging all the stakeholders. There are different KPIs as per the job functions. Next step is to focus on analysis and visualization as per the relevance and impact.

This step is all about clarity and focus. Before diving into the data, I ensure we have a solid understanding of our objectives, such as enhancing operational efficiency or reducing resource waste. This involves selecting specific, measurable metrics like OEE for productivity or MTBF (Mean Time Between Failures) for equipment reliability. This upfront clarity streamlines the analysis, focusing efforts on data that directly impact our targets.

In defining the goals and objectives, it is also important to recognize if the KPI’s are tactical or strategic. Strategic KPI’s need to be linked to executive level initiatives. In the process of managing your CI, there will be the need to temporarily implement tactical data collection for analysis. This may require relinquishing the strategic KPI’s for a short term but needs to be controlled and managed properly.

The key technique before visualization of MES Data is data contextualization. MES brings together data from multiple data sources and formats from multiple layers in the ISA-95: Level 1 Controls Level 2 Supervisory Level 3 Manufacturing Execution Level 4 Business Planning and logistics It is important to clean, aggregate and then contextualize the data before trying to visualize. There are typically two types of MES data that will provide "context": 1. Time series data - e.g. equipment reporting pressure measurement every second. 2. Transactional data - e.g. measurement reported against a unit/lot/batch We need to be able to marry the two. For example, What batch was running on the equipment when the pressure dropped?

As part of the determination of the right format, it helps to recognize whether the data is being collected as part of a normalized process (standard production process) or an altered process for analysis and investigation. In a normalized process it is possible to aggregate several data elements similar to OEE or a bell curve graph. This allows tracking of several aspects of the process in one KPI that is good for process indicators. In a process being used for analysis, the trending of raw data is more important.

As part of the visualization, it is important to ensure that the visualization makes sense to operational staff. I have seen far too many time, the implementation of graphs and other data presentations that were in place to provide visual context but to an operations person, the presentation did not make sense, as in showing multiple data trends on a single graph when the data trends had no relation to each other operationally.

The landscape of data analysis tools is vast, but my focus has always been on those offering robust integration with MES systems and the ability to process data in real-time. Tools like SAP HANA for its in-memory computing and Tableau for its advanced visualization capabilities have been indispensable. Furthermore, adopting machine learning algorithms for predictive maintenance has significantly reduced downtime, making our operations more reliable.

Data visualisation is crucial in conveying insights from MES data effectively. To ensure clarity and accuracy, choose the right chart type, utilise clear labelling and contrasting colours, and avoid clutter or distortion. For example, pie charts can look nice, but it's not often that they are the right chart for the information. Remember, the goal is to facilitate understanding, not to obscure it.

Visualization is where data speaks to us. I apply principles such as choosing the right chart type for the data in question, using colors judiciously to avoid confusion, and minimizing clutter to keep the focus on the insights. For instance, using a time series graph to track production volume over time or a heatmap to identify bottlenecks in the production line. This not only aids in quicker comprehension but also facilitates more informed decision-making.

Adhering to data visualization principles ensures your data is presented clearly and effectively. Principles include clarity, accuracy, and efficiency. Use appropriate chart types (e.g., bar charts for categorical data, line charts for trends over time) and avoid clutter. Ensure your visualizations are easy to interpret, using consistent colors, labels, and scales. Always aim for simplicity without sacrificing necessary detail.

It's crucial to ensure that the data is providing the desired outcome for the users. Seek their feedback, and remember, it's not going to be right first time - it's ok to keep refining. It's crucial to choose a flexible MES that allows users to create their own reports and dashboards, giving the freedom to iterate.

This is a continuous cycle of improvement. Initially, not all analyses yield actionable insights. I employ a cycle of feedback, particularly from those on the production floor, to refine our analysis and visualizations. This might involve adjusting the data collected, changing the analysis parameters, or even revisiting the goals. This iterative process ensures that our analyses remain relevant and impactful.

Data analysis and visualization should be iterative. Start with initial visualizations and analyses, then gather feedback from stakeholders. Look for patterns or insights that may not have been apparent initially and refine your approach accordingly. This may involve cleaning the data further, adjusting visualizations for better clarity, or exploring additional data points. Testing and refining help ensure the final output is both accurate and useful.

Reporting is such a key part of MES, beware of MES applications that collect the data, then rely on other separate applications for reporting - this is a recipe for disaster.

Quality of Data: I place a high premium on data quality, investing time in cleaning and preprocessing data to ensure its accuracy. This might involve dealing with missing values, correcting errors, and ensuring consistency across data sets. Cross-Departmental Collaboration: The insights gleaned from MES data analysis are shared widely, not just within the production team but across the organization. This collaborative approach ensures that decisions are informed by a comprehensive view of the operational landscape, fostering a culture of continuous improvement.

Additional considerations might include data security, user training, and scalability. Ensure data privacy and security measures are in place to protect sensitive information. Train users on how to interact with and interpret the visualizations to ensure they can effectively leverage the insights. Lastly, consider the scalability of your solutions to handle increasing amounts of data and more complex analyses as your operations grow.