Many manufacturers have implemented costly ERP (Enterprise Resource Planning) systems in order to track manufacturing KPIs and have tried to leverage those systems to provide real-time data from the factory floor. While not purpose-built for this, most ERP systems are remarkably flexible (notice we didn’t say agile), and data can technically be streamed from the floor, transformed, put into some intelligible context, and displayed within an ERP system, but it fails to impact productivity in a number of ways. Feel free to read earlier blogs on the subject: Manufacturing ERP Systems Don’t Do What You Think They Do.
Part of the issue is that the manufacturing industry at large, from operations to marketing to media, often lumps the kind of connectivity, monitoring, and plant performance analytics built into a product like Mingo with the broad, business-focused data management and intelligence promised by an ERP, BI (Business Intelligence) or MES (Manufacturing Execution System). This is a mistake and hides a glaring analytics gap.
The “Big” Picture of Manufacturing Business Intelligence
Take for example a recent report on the market potential for Manufacturing Analytics: Manufacturing Analytics Market worth 8.45 Billion USD by 2021. The report goes on to mention big players in the space, e.g. Tableau, Microsoft’s Power BI, IBM, Tibco, Sisense; without mentioning that some of these are general analytics platforms, not specialized for manufacturing, which means that dashboards and KPIs in these systems are meant for management in the back-office, not the plant floor.
The confusion in the marketplace becomes even more apparent when you look at the number of reports on IoT adoption which commit the same error. This can be illustrated in a recent report by Plex which identifies the gap, but unfortunately doesn’t provide any numbers.
“While many manufacturers are already using analytics for management insight (78 percent), there still seems to be a lag in using shop floor data to improve operations…The huge opportunity is to really leverage the data coming from the shop floor to give insight into operations.” Source: 4th Annual State of Manufacturing Technology Report
Anecdotally, we see this all the time. Manufacturers have more data and analytics than ever before, but they still lack visibility into the root causes of downtime, inefficiency, and quality issues. In a sense, there are two different realities at play, two different tales — The one on a business manager’s laptop and the realities on the plant floor.
To help clear up this misconception, and hopefully bridge the gap, we’ll outline the difference between the two types of “manufacturing” analytics and try to make a clear distinction between the two.
- Operations or Enterprise Manufacturing Analytics are those focusing primarily around business operations, e.g. supply chain management, financial forecasting, asset management, top-line or summary productivity data, and predictive analytics through human or machine-learning data analysis to provide an operational view of the company and its financial inputs and outputs.
- Machine-based, Plant Floor Performance or Productivity Manufacturing Analytics on the other hand aggregates data from machines and machine operators and provides dashboards and monitoring solutions specifically to provide visibility into daily operations on the plant floor.
The two are very different. In the first case, data is aggregated from a variety of different systems and inputs, usually, as a batch process, is rarely real-time, and isn’t particularly well-suited for troubleshooting or root cause analysis on the plant floor. Operations or enterprise manufacturing analytics data is housed in either an ERP (Enterprise Resource Planning), a Business Intelligence platform, MES (Manufacturing Execution System), or MOM (Manufacturing Operations Management) and is primarily driven by historical data to produce specific reports mostly for business analysis. Most machine data that might be useful requires an edge server and little of it makes its way back to the floor for real-time adjustments. The data collected may or may not be stored in an operations database or a historian.
In the case of a product like Mingo, data is captured specifically to track machine, operator, and process performance, and is purpose-built for productivity and performance to be used in real-time or for trend analysis of events on the factory floor. It is primarily a data collection tool, a monitoring solution for the plant, deptartment, or cell and machine-level intelligence. Data is gathered to help calculate availability (downtime), measure performance (throughput) and quality (scrap or yield), and in some cases condition monitoring (temperature, vibration, etc.)
For a plant floor manager, a dashboard in an ERP is most likely not created for this purpose, meaning it lacks machine, part, or shift-level context and may be hard to access from the plant floor, while Mingo can be easily configured by department, cell, machine, etc. and is made to be accessible by role and department, on a tablet, from anywhere. It is purpose-built to help provide machine performance and plant productivity gains.
Perhaps we could avoid the confusion by always including “machine” with the term manufacturing analytics, but this isn’t even strictly the case. Mingo allows manufacturers the capability to take input from operators to provide supplemental context around jobs, downtime, and quality, for example, providing much more than just annotations, it provides additional data points. Without this functionality, a big part of the real performance picture is missing. Plus, we have customers that use Mingo solely to take operator input from our operator UI, since a great deal of value is in the contextual and real-time display of data, automated or otherwise.
Maybe we should start calling what Mingo does “plant productivity analytics” or “plant performance analytics” to help differentiate it. Hopefully, though, this distinction becomes a moot point since in the future we expect to see productivity analytics technologies bundled together with popular manufacturing systems or implemented as part of an industrial analytics solution stack.
Most analysts agree that real-time plant floor analytics will help drive IoT adoption, especially as potential ROI increases as part of an overall predictive maintenance and downtime reduction program. Plus, since the promises of Big Data and AI through predictive analytics require lots of this type of data to be of any use, downstream business intelligence applications will be limited without it. For now, it’s clear manufacturers need to invest in productivity solutions that will work at the plant level, and embrace the fact that all continuous improvement efforts, i.e. Lean, TPM, Six Sigma or any other methodology, always start on the factory floor.
