Process and Data Automation for the Mid-Sized Foundry

Jim Lagrant

What is an automated foundry going to look like in the future? Do you foresee additive manufacturing cells and collaborative robotics seamlessly interconnected and providing big data to advanced analytics? For most foundries with fewer than 100 employees, probably no time soon.

The gap between operations as they are today and an interconnected, automated enterprise is more like a chasm in terms of finances, skillsets, and physical limitations. However, like many other manufacturers, metalcasters are beginning to realize that having a connected enterprise to see how their process is performing, and making better decisions more quickly, is what will enable them to keep their company competitive. It can be intimidating to think you have to take a plunge into Industry 4.0. But if you consider this digital transformation as a journey from understanding to advancing to outperforming, you’ll see that there are incremental steps you can take to start gaining knowledge about your process without breaking your budget.

Several solutions and devices are available that are scaled and applicable to the metalcasting industry. 

Foundry equipment is typically overdesigned and  built to last. For instance, it’s not unusual to see sand mixers and breakdown furnaces installed in the 90s still in use. This equipment continues to run day after day, but it doesn’t provide a lot of information and it isn’t suited to the people you are hiring today to run or maintain them. 

Some other equipment in a foundry may be a little newer and is only a couple steps and a very small investment away from producing useful information.  This automation can improve your operations and lessen some of the impact of an aging workforce. Applying process data collection and analysis of business metrics gives all levels of the organization a common view of the health of the business.

In metalcasting, a lot of hands are involved in the process. The equipment is not a frontline player as much as it is in other manufacturing plants such as packaging or assembly. In a foundry, equipment issues tend not to be noticed by engineering or management because operators are good at working around issues that crop up. Due to the manual nature of foundry operations, these issues­—and workarounds—can be unrepeatable, which leads variability in our product. 

Because of this variability, there is a lot to be gained in automating a manual process. You might begin at looking at the pain point or the problem to be solved. Do you know how much energy it takes to hold your metal overnight, or how often your heaters fail? That might be reason enough to collect data from  the process. Is your melt quality degrading due to inconsistencies in additions? Maybe you want to add visualizations and prompts so the manual steps are being followed consistently.

If you are having metal-to-mold reaction, do you know what your binder ratios are, or the last time they were changed? If you are out of capacity, do you know when your equipment is running or is idle? You might be surprised to see how often your machine is not running during the day. Or, do you get complaints from your operators that it is too hot, too cold, or the air quality is bad because the exhaust system was turned on or forgotten to be turned off? Offloading some of the standard actions that operators do, such as process timing or auxiliary systems such as ventilation, can allow them to focus on what’s important: making good product.

These improvements can also lead to improved operator effectiveness, better decision-making, and fewer errors by providing them with added features, such as human/machine interfaces, alarms, and trend screens that give access to information and recommendations for action. It will also help them adopt new practices, procedures, or workflows that will improve their productivity. 

The application of process sand data automation was applied at a mid-sized aluminum sand foundry with high mix, low volume work. Now the foundry employs multiple connected systems that collect, analyze, and report on the factory. Time series data is collected around the clock from its equipment. Visualization was added at key processes. Information is displayed to operators and they are allowed to manipulate the process in a controlled manner. Operators are restricted in what parameters they can adjust as well as how much adjustment they can make. 

The foundry also added the means of collecting transactional data such as melt temperature at the pouring operation. That might seem obvious, but once the data was available electronically, it could be fed into analytical software for charting and alarming.Data that is entered into terminals on the floor is fed into statistical process control software which tells staff whether the process is in control or out of specification, and a top level dashboard for foundry supervisors allowing them to see right away which part number, for instance, has an issue.

The foundry achieved its system slowly and in stages and by managing its capital. 

Getting Started

To begin implementing process and data automation in your own metalcasting facility, first verify if the piece of equipment you are considering is capable of meeting the needs of the customer. Customer demands might have changed, or new environmental regulations might have passed since it was installed that the equipment may be incapable of meeting. In this case, your money may be better spent elsewhere.

Once you’ve identified the process, it doesn’t have to be expensive or involved to start collecting data to characterize how it is running. A small investment in a USB data logger and a motor current transmitter (less than $500) can provide valuable information. You will be able to find out cycle time, machine loading, order of operations, and percent utilization. After you review that data you can make the decision if you need to get more sophisticated on your data collection project.
One of the big challenges in automating manual processes is the real or perceived lack of repeatability. Even though it is a manual process, at one point in time, there were a series of tasks that were determined to be the best method. Over time, those manual processes tend to degrade. The degradation can be worse if best methods were not documented. It is imperative that you get agreement between the operators and subject matter experts on the correct sequence of steps and what-if scenarios. Once that is done you can begin documenting the functional specifications of the automated system used to create the data collection scheme, control scheme, and revised work instructions.

Upgrading Equipment

There are three approaches for equipment upgrades. 

  • Best in Class: Retrofit existing equipment with third-party connectivity solutions and add sensors that directly measure key process indicators.
  • Rip and replace: Fully scrap legacy equipment and replace it with modern, communication-enabled machines.
  • In-house solutions: Solutions created by internal personnel and technical resources. These are infinitely customizable to pinpoint and address known issues.

One example of retrofitting existing equipment is a nobake sand mixer that has a legacy PLC from the OEM. For a small investment, a communication module was added to the PLC to make all the data available from that process. Now that the data was available, it needed to be stored somewhere more sophisticated than an Excel spreadsheet. In the past, this data was typically stored onsite in a process data historian server. Today, there are numerous cloud data storage options that can be selected a-la-carte, and do not require significant internal IT resources. Therefore, if you have equipment with controllers in them now, there is no reason to not collect data from them.

In a rip-and-replace scenario, you are replacing outdated equipment with the most up-to-date technology. For instance, another mold mixer that was installed in 1992 was working fine from a mechanical systems standpoint but there was no visibility in the process. The foundry could not tell how frequently the binder ratios were being adjusted by the operators on the floor. For less than $20,000, new monitoring and control equipment was built up and tested offline before being installed during maintenance shutdown.

In-house solutions can help address the needs of new hires. If you look at the new employee who is running your equipment today, they  typically do not have a manufacturing background. Their previous experience might be from Dunkin’ Donuts and McDonald’s. These employees are used to having access to information at their fingertips, and they are eager to learn. Process visualization delivers job specific instructions, enforces rules and reduces the complexity for the inexperienced worker.

The aluminum sand foundry did this by expanding the use of existing equipment. For example, it added  communication cards to existing temperature controllers and tied them into the melt deck supervisory touchscreen. This way, the melt deck operator can manage the temperatures remotely. Logic was added to bring the furnaces up to temperature overnight to limit electric peak demand. Without this logic, the furnaces were all turned up at the start of the shift, resulting in a large peak demand charge. This is a good example of an in-house solution that automated some of those overhead decisions the operators have to make.

Automating and Analyzing Transactional Data

An issue with automating business transactional data is that it comes from multiple data sources (Excel, access, ERP), and most mid-size foundries have limited internal IT support and a small operating team.

In this scenario, working with a third party to help create your platform may help bridge the gap. An analytics platform can enable you to connect your multiple data sources, get real-time, hands-off analytics, role-based dashboards, and alarming and notification services for extra attention when a process drifts out of specification.

User-specific dashboards lead to KPI charts for statistical process control (SPC). One of the questions that is often hard to answer is whether a process is doing better or worse than last month and last year. Differing time spans display immediate data as well as long-term trends, giving a clearer vision of how the business is doing. Since the dashboards and charts are automatically updated as soon as raw data is generated, all of the time spent on scrubbing the data and generating the graphics is eliminated. High level dashboards indicate business performance, and consistency in graphics significantly reduce the time spent debating what you are looking at.

In the past decade there has been an explosion of connected consumer technology. Think of how many people you know that have fitness trackers or smart watches telling them their heartrate or providing directions. This commercial technology has been hardened for industrial use, resulting in a great number of connected industrial devices that can be added to equipment to collect data and integrate with existing business systems. This data can be integrated into handheld devices to display data real-time where it had not been previously available. These handheld devices can run augmented reality (AR) platforms for equipment training, maintenance, and operation. These platforms are one more method manufacturers can use to close the workforce labor and skills gap.   

This article is based on an AFS webinar originally presented in July 2019. AFS members can watch the full recorded webinar at www.afsinc.org/members-only-webinars. Jim Lagrant was the Vice President of Engineering at Palmer Foundry in Palmer, Massachusetts. He is currently the director of the Manufacturing Engineering program in the Mechanical and Industrial Engineering Department at the University of Massachusetts-Amherst. Jim is seeking opportunities for students to work with manufacturers, OEMs and suppliers who are interested in the transformation to Industry 4.0. For more information, please contact him at jlagrant@umass.edu.

Click here to see the article in the April 2020 Modern Casting digital edition.