Capturing Data from Manufacturing Equipment, Old and New

By Mary L. Martialay (RPI)

RPI MillRensselaer Polytechnic Institute hosts a Smart Manufacturing Innovation Center (SMIC), a regional extension of the national Manufacturing USA and Department of Energy-supported smart manufacturing initiative known as CESMII. The CESMII SMIC at Rensselaer is currently engaged in a project with industry partner Toward Zero to collect operational manufacturing process data from a broad variety of manufacturing equipment and make it available by way of a single seamless interface. Working together, the team has applied new technology from Toward Zero to capture data from legacy machinery in a way that will prove easy and repeatable for commercial manufacturers. In this post, we hear from Craig Dory, director of the SMIC.

Q: Explain what you are trying to do in this project.

We are collecting operational data from a broad range of manufacturing systems, regardless of the age, type, or system builder, into the CESMII Smart Manufacturing Innovation Platform (SMIP). Our smart manufacturing partner Toward Zero is helping us connect to and collect data from a wide variety of manufacturing equipment, providing a direct connection to the CESMII SMIP via GraphQL which eliminates the need for third-party software and dramatically reduces the cost of connecting to the CESMII SMIP. The goal is to demonstrate how we can make manufacturing process data available from different manufacturing systems across the enterprise by way of a common and accessible data structure. Working with the Rensselaer Manufacturing Innovation and Learning Lab (MILL), and the Chemical Engineering Department, we’re currently accessing and normalizing data from machinery housed here at Rensselaer, including lathes, CNC machines, plastic injection molders, water jets, and 3D printers. The variety mimics what you might find on the shop floor of a small to medium-sized manufacturer, a mix of older and newer machinery from many different manufacturers.

Q: You have an example of that, right? Tell me about your most recent success.

We recently began capturing data from a HAAS ST-10Y CNC lathe in the MILL using a prototype smart manufacturing edge appliance developed by Toward Zero, called Apogean— It connects directly to the ST-10Y to capture the process data stream from the metal parts it cuts. Apogean then converts this real-time data and instantly outputs it following a number of industry-standard protocols including OPC-UA, MTConnect, and CESMII’s preferred GraphQL, to facilitate a flow of complete, contextual, quick, and accurate data into CESMII’s new Smart Manufacturing Innovation Platform. This is the first-ever use of such an edge appliance. Going forward, Apogean will be used to connect to a broad range of modern and legacy manufacturing systems to capture and contextualize manufacturing process data.

Q: What’s the benefit of capturing data from manufacturing machines?

With contextualized data about their manufacturing systems, manufacturers can make better decisions about how they employ their manufacturing processes. Here’s a simple example: you can improve overall productivity by understanding some of the factors that cause things like tool wear, so that you’re proactive in selecting when you want to do maintenance on your machines, or even optimize processes to minimize tool wear so your tools last longer. In many respects, access to contextualized data on what your machines are doing is the starting point for making smart decisions about your manufacturing processes. Once we can effectively monitor machinery by collecting rich operational data, we can also model their performance and predict outcomes, even develop model-predictive control for “lights out” operations.

Q: Is it difficult to capture data from manufacturing machines?

It depends on the machinery. Just looking at our shop, we have equipment ranging from 40-50 year-old legacy lathes that have little or no built-in sensors or data-collection abilities, to completely modern CNC machines that are computer controlled and have access to time-series data from a number of operating parameters that can be downloaded with ease.

For older machines that don’t have any computer control, or don’t have any sensors, we have to sensor-ize that equipment – deciding what types of information that we want to draw from those manufacturing systems, installing sensors, capturing signals, wrapping context around the data, and off-loading the data streams from those machines to apps, data lakes, or data warehouses so humans can use it for decision making. Apogean lets us add sensors to any equipment regardless of age – as an example, we can add power or temperature sensors to a 50-year-old machine to deliver valuable data to develop energy-saving strategies, or create an optimized preventive maintenance plan.

With the modern systems, we can literally plug into them and download operational data, but the difficulty is that most system builders employ their own proprietary data protocols. That may make business sense for them, but it can make it difficult for us to get operational manufacturing data standardized into a common data structure. In those cases, the real challenge is to interpret the machine protocol, the language that those machines talk, to be able to download the operational data. If we can do that, we can take the data that’s made available from the manufacturing systems and create data streams with a standard data structure that can be stored in the cloud and brokered to various uses such as production dashboards or applications for deep data analysis. In that way, a manufacturer can access its manufacturing process data from across the enterprise in a consistent form to enable better decision making.

Q: What kind of data can you capture?

With Apogean, there’s a huge breadth of information that we can capture and contextualize , something we couldn’t quickly, easily, or cost-efficiently do previously. From the modern systems, we can gather documentation information like controller information, machine names, serial numbers, who the operator is, the software version, and production information like part counts, program data, and name. Then we can look at events such as alarm codes and feed rate overrides, and different types of codes that describe the actual operation of those machines. We can also accurately capture the required inputs for overall equipment effectiveness(OEE), plus part counts, tool positions, spindle RPM, and state of machine and tools in use, which is critical analytical information we need from both modern and legacy machines. A high OEE means that you’re manufacturing all good parts with no unplanned downtime, which is the gold standard. It’s hard to achieve, but helping manufacturers reach that standard is our goal.

This fall, the Rensselaer SMIC plans to hold an open house illustrating the value of smart manufacturing and showcasing CESMII’s Smart Manufacturing Innovation Platform. Stay tuned for a date and further details.