Using incremental process control for largely manual processes
Some food and beverage plants—which may more closely resemble chemical or refining facilities—are potentially well-suited to totally automated process control. Examples are plants for yogurt, milk drying, edible oils, certain candies, soda and breweries.
However, other applications like ready-to-eat sandwiches, mixed salads or custom soups—especially largely manual processes—may not be so easy to automate but can still benefit from incremental process control implementations.
While process control strategies and their definitions (e.g., PID, feed-forward, model-predicative, etc.) don’t change over time, the technology to implement these strategies has come a long way in the last 20 years, becoming more transparent to operators and more powerful in the knowledge provided to businesses. The system you need is within reach, but long-term planning is key. Without planning, you could wind up with a system that provides little actionable information and is next to impossible to upgrade.
The scope of existing systems
Sometimes it’s easy to spot when an existing system has been rubber-banded together to such an extent that it’s not even practical to consider upgrading. And in some cases, the problem is not only with the control system, but also with the plant equipment and the building housing it. Dan UpDyke, market development manager for CPG and life sciences industries at Rockwell Automation, outlines just such a case.
A french fry manufacturer’s existing facility had seen better times. It had an automation system that had been pieced together over the years and had a great deal of custom engineering. It was hard to maintain and modify to accommodate production changes. The company also needed to integrate the system with packaging equipment, subsystem skids, CIP and filtration systems. In this case, the best solution was to start with a clean slate—a new facility.
In the new building, a Rockwell PlantPAx distributed control system was installed and provided the plant-wide control and flexibility needed for the operation to be successful. The entire system was simulated for factory acceptance testing (FAT) to reduce commissioning time. The processor saved hundreds of engineering hours with standard code and tools, and this helped to reduce training and startup times.
In the end, time to market was reduced, and this processor now has a system that is easy to maintain and flexible enough to handle future manufacturing needs. The facility served as a pilot project, and the PlantPAx DCS is now the standard for the company’s french fry production operations.
Sometimes a relatively simple controls upgrade can result in a major improvement. Matt Scott, lead systems integrator/director of mentorship at Malisko Engineering, a Control System Integrators Association (CSIA) member, describes one situation. A completely manual mixer operation, requiring many precise raw material additions and maintenance of several critical control points, was converted to an automated batching system.
“Using off-the-shelf batching software to control extremely standardized control phases saved huge amounts of capital in the form of labor cost, raw material usage and batch turnaround time,” says Scott. During this evaluation process, it became clear that the way the operators were making product was far from the expectations of engineering and quality. “Through spending a lot of time with operations and monitoring their process, we were able to optimize so many procedures. We took procedures that were done five different ways by five different people and made them the same way and the best way every single time.”
What were the benefits? “Operator satisfaction, batch consistency, standardization, massive cost savings and extremely fast ROI,” says Scott. And what any SI wants to achieve: great customer relationships and repeat business.
Here’s another situation: A company wants to keep the facility and add new equipment and software. That’s what Prima Frutta was able to do when it installed new machines and Ignition software, says Kevin McClusky, Inductive Automation co-director of sales engineering. The software provides visualization on 120 screens around the facility.
“The biggest stumbling block to data collection is often device connectivity and data availability,” says McClusky. “If you start with a tag list from your OEMs and PLC addresses, then Ignition is normally able to connect directly and gather process data, exchange recipes and provide visualization and reporting on your processes. A tag list and/or access to the PLC program from the OEM can be critical. Writing these things into the requirements when you’re shopping for a new set of equipment can make data collection later much easier.”
Different companies receive different gains, adds McClusky. In the Prima Frutta case, production increased 50 percent without expanding the workforce. For another client, SugarCreek, which produces raw and fully cooked food products, Ignition enables easy access to real-time data and provides information in a context and format that can be used to improve the business. (SugarCreek was FE’s 2016 Plant of the Year.) For others, the software provides scalability, access to data while in any location, alerts for quick response and tuning recommendations to run more efficiently, says McClusky.
Ancient PLCs and software is one issue that Chris Schulze, business development manager at Advantech, a CSIA partner, often sees.
“Our recommendation is to replace old, outdated PLC systems (which can’t handle the influx of new data and new systems) with compact, fast and less expensive industrial PCs (IPCs) utilizing a soft PLC, such as CODESYS,” Schulze advises. “Common processor struggles we see include obsolete factory and company automation standards. These standards were written 30 years ago and are often never revised.”
Updating controllers brings various advantages, including better product quality, more output with the same machines, less waste, tracking and traceability, better visibility of production, higher OEE and risk mitigation, he says.
Updating mostly manual systems
Mostly manual operations can include several applications, such as weighing and batching, making sandwiches or RTE salads, or producing custom, made-to-order kettles of soup. While some processes may be impossible to automate completely, automation of parts of the process may prove beneficial.
“Consistency is essential for a food and beverage producer’s bottom line. Even with manual assembly processes, control-like systems can be applied to maximize product consistency,” says John Fryer, Stratus Technologies senior director of industry solutions. For example, in the case of RTE sandwiches, processes such as ingredient weighing and portioning can be automated to support manual assembly. These systems also could track product ingredient use to ensure consistency, rather than just meeting the final weight of the product with a more arbitrary use of ingredients. The same can apply to temperature and quality standards, adds Fryer.
“It is very important to assess the current process and discuss where the pain points are,” says Malisko’s Scott. From there, provide recommendations to fix them, starting from the most critical. Make sure to address the operation from the plant floor up and assure the best control methods are used. A statistical process control (SPC) system can help assess problems.
“When applying a solution to a mostly manual process, our application engineers prefer to map it out using a design of experiments (DoE) methodology,” says Garrick Reichert, Opto 22 applications engineer. This helps to capture and log all the independent and control variables, establishing practices that are reliable and repeatable—all while being maintainable along with the manual aspect of the operation.
“Once the process is in place, you can implement a continual SPC element to maintain the operation using the DoE data as a template for your control limits and variation management,” adds Reichert.
“Since we sell software, if a solution includes hardware, our discussion is generally with an integrator,” says Inductive Automation’s McClusky. For mostly manual operations, Inductive Automation’s integrators often recommend a few easy additions:
- Simple screens or tablets to give operators a sense of progress through work orders and allow them to fill out info in the system, rather than using paper forms
- A simple sensor or two and lightweight Ethernet-connected data collection units (such as the Advantech Adam, Moxa ioLogik, etc.)
- A panel (such as Ignition Edge on a Panel PC) to do some local data collection, using built-in store-and-forward for data resiliency
Food processing and packaging tend to be a low-margin business, so selecting a PLC with a high price/performance ratio is critical, says Don Pham, IDEC senior product marketing manager. The selected PLC must be easy to use for programming for initial design and ongoing support. This is very important as many food plants don’t have a staff of engineers on hand to program PLCs. In many cases, one person is expected to support multiple PLCs, often across multiple facilities, making remote access a necessity. Look for Bluetooth or Wi-Fi connections for local programming access for PLCs.
For some applications, new architectures can help. Wireless capabilities, cloud architectures, 5G networks and low-power wide-area networks (LPWANs) will greatly influence a processor’s new system and its data flow, says Benjamin Jude, global solutions architect for Schneider Electric, industrial automation-consumer packaged goods segment. However, network reliability and cybersecurity are certainly two major criteria that cannot be ignored when implementing a stable and sustainable digital infrastructure.
Speaking of new architectures for automating manual applications, Advantech’s Schulze sees some potential solutions, such as more agile automation. For example, pick-and-place and specialty robotic solutions are becoming less expensive for several types of operations.
“But also remember, when you’re replacing a manual process where the operation is impractical or not cost effective, improvements can still be made by applying IIoT technologies within ‘error-proofing’ applications,” adds Schulze. “For example, electronic work instructions enhanced by augmented reality are gaining a lot of interest in the industry.”
Today, the trend in control systems is to use open standards wherever possible, which simplifies design and improves connectivity. “The most important consideration for selecting a process control system is the level at which open technologies are embraced,” says Schulze. “There is no one-size-fits-all control system, so it is imperative that the system selection includes the openness necessary to easily integrate other technologies and data flows.”
Control strategy is one of the first things to establish in order to determine a direction on automation, says Malisko’s Scott. Does the process perform the same functions every time? Is the process continuous, or does it have a defined start and finish? In many cases, the control strategy can be transparent to the operator, but on-site resources often are required to maintain a control system, depending on its size and complexity. An example is recipe control and creation.
“Keeping any given strategy simple means that you can rely on less technically capable personnel to reliably operate and maintain a given system,” says Opto 22’s Reichert. But processors can and should hide the strategy from operations as they typically do not need control strategy expertise.
Instead, operations should be given well-documented standard operating procedures (SOPs), whereas maintenance and technical personnel often need detailed turnover packages (TOPs) showing P&IDs and loop diagrams, along with documenting control constraints and definitions.
Multiple considerations affect selection of a process control solution, says Melissa Topp, ICONICS senior director of global marketing. An immediate factor to consider is existing equipment compatibility, both for production devices such as mixers, extruders and ovens and for PLCs, sensors and other automation hardware.
Another area is compatibility with existing infrastructure, such as vital network communications protocols, including OPC UA, Modbus, BACnet, SNMP, web services, etc.
Organizations will need to evaluate how their selected process control system will tie into other enterprise-wide production-related systems, such as ERP or MES, adds Topp. Processors have to figure out whether the application will require IIoT connectivity and whether the process-based equipment is IIoT ready or will require retrofitting or the use of edge devices (e.g., IIoT gateways).
In a batch process environment, the type of process (heating, baking, freeze-drying, etc.) usually defines the process control system to be implemented, says Matt Wells, GE digital VP of product management. However, before deciding on a system, processors usually take into account the overall dynamics of their operations, such as type of products, production schedules, food safety requirements and processing equipment availability.
For a complex processing operation that manufactures multiple products and requires stringent cleaning procedures for food safety, processors usually combine process control strategies with batch management strategies. In those cases, control platforms that can integrate into batch management concepts are generally preferred to ensure a consistent interface to controlling operations, adds Wells.
Sometimes a very important consideration gets overlooked in planning control systems, Fryer of Stratus Technologies says. When selecting process control systems, like any technology, processors should carefully assess their risk tolerance for equipment failure. For some, the impact may simply be decreased production. For others, unplanned downtime could prove catastrophic to food safety.
Businesses with low risk tolerance and low technological expertise should prioritize systems with little to no downtime. They should also establish a clear line of ownership for control system failure and a plan for fixing problems.
Other processors can consider relying on third-party system integrators, advises Fryer. These integrators can sometimes fix software issues remotely. Hardware problems, meanwhile, require integrators to visit sites and manually repair technology.
That is one reason continuously available systems are essential for some operations, with internal, on-site talent to address system failure. Essentially, the less sophisticated a plant is about controls, the more important availability becomes.
Test, test and test again to help eliminate any “gotchas” and minimize risk.
“If an end user is buying a machine, it should be fully tested by the OEM machine builder before shipment, with the testing witnessed by the end user,” says IDEC’s Pham. “Once installed, the machine should be tested again, this time simulating actual operation to the greatest extent possible. Both testing regimes should include training for end-user personnel.”
Plan supporting architecture carefully
While modern food processing equipment often has built-in controls, a supervisory control system makes sense when there is a need to automatically sequence operations between different systems and/or tie in additional systems to the control of the process, says GE Digital’s Wells. Additionally, when there is a need to standardize the operating interface for operator effectiveness (e.g., training, easing the switch between lines, etc.), leveraging a supervisory control system can be beneficial.
Planning ahead to avoid data logjams is a good idea. “We would recommend that processors don’t attempt to tie every system into one large connected pyramid of devices,” says Advantech’s Schulze. That approach almost always results in a lot of data being collected with no plan for analysis. A better approach would be to look at the specific needs of the manufacturing operations and then develop the applications to meet those requirements. The solution(s) may be a mixture of purchased and homegrown applications.
“Our preference is to avoid layering too many different systems to operate your plant,” says Opto 22’s Reichert. With most operations these days, there is a push to flatten the process, reduce pieces and parts, and make the system as simple and straightforward as possible—all while addressing cybersecurity.
“Now that edge-of-network devices, like edge programmable industrial controllers (EPICS), have built-in elements of a supervisory system, and the supervisory system may have a fully built-in MES-type environment, finding simple, elegant solutions to traditionally complicated problems is becoming more the norm,” says Reichert.
“In this area we have made great progress using secondary sensing technology with wireless sensors and even batteryless sensors,” says Schneider’s Jude. Zigbee, Lora, SigFox and all IIoT boxes or gateways allow the connection of old machines to software with limited cost. As an example, a magnetic sensor without any wiring or cables can measure temperature or humidity and push that data to a cloud-based application using a Zigbee gateway. Or a nonintrusive sensor can measure the energy consumed by a machine and push the data to a cloud application without touching the machine or the PLC code.
“With a couple of hours of work, your old and nonconnected machine can begin sending data to a cloud-based application where you can start running analytics, opening a new world of opportunity,” says Jude.
Do you need a MES? It depends. “Modern HMIs have a tremendous range of capabilities, so traditional MES applications are usually not required,” says IDEC’s Pham. “The right HMI should be able to handle all but the most sophisticated tasks at lower cost and with improved ease of use as compared to an MES solution.”
“Connecting systems to a larger control system almost always makes sense, whether this is a supervisory control system (SCADA) or an MES,” says Inductive Automation’s McClusky. “If you do this, you can monitor your processes centrally and collect data on performance. Performance data is critical in order to properly assign resources for maintenance, tuning, upgrades or capacity planning.”
If built-in controls are used, tying them into related systems may be beneficial, says Malisko’s Scott. Systems for raw materials, filling, packaging and CIP can all benefit from getting pertinent information from each other to form a complete and intuitive system. If a sensor or system asset is on a network, it is becoming important to expose that information, so its attributes can be utilized.
Integration concerns now and in the future
Many food and beverage enterprises will most likely have pre-existing ERP or MES implementations and should seek process control software that can integrate with both, says ICONICS’ Topp. For quality assurance, these organizations should seek a process control solution that helps monitor and manage quality and provides native graphs and trends. Operators, quality personnel, manufacturing engineers and management should be able to view SPC data and other production parameters that affect product quality, adds Topp.
Operations and quality personnel need the ability to see their data on any platform whenever they wish, without demystifying unintelligible tag/data point names, says Opto 22’s Reichert. That way they can organize the information, interpret it and act upon it without wasting time hunting for it.
“The key to integration is the extent to which the process control system is open with connectivity technologies,” says Schulze. “The more options, the better, but there will always be challenges to achieve business goals. We are seeing more and more that MES, LIMS and Quality will be integrated in the soft PLC on the industrial PC.”
“Almost all control systems ‘can’ connect to external software systems, and a salesperson from those companies may make it sound easy, but dig a little deeper to get a true sense of effort,” says McClusky. “Integration projects with software systems can easily be one of the highest costs of implementation, or they can be simple and relatively inexpensive if you’re using the right software. It’s also worth noting that some software can be installed alongside of existing systems to ease the effort required if your existing process control system is time consuming to integrate.”
“With any system, plant supervisors must consider how well they can seamlessly integrate new systems into their existing process,” says Tanner Sullivan, Cat Squared technician. “The point of integrating software solutions is to enhance current processes, and if a potential solution creates excess, redundant work, it is no longer an improvement to your process, but it becomes a hardship.”
Finally, remember that process control systems are an investment for the future, says Stratus’ Fryer. Choose control systems that are open and fully conforming to modern standards. Closed process control systems are extremely risky investments, since businesses, especially as they scale, can rarely predict their technological needs.
“Say a beverage producer doesn’t have a laboratory information management system (LIMS) right now but plans to install one within the next few years,” says Fryer. “Choosing a process control system that will be compatible with a LIMS in the future is essential to support long-run connectivity.”
For more information:
Cat Squared, www.catsquared.com
GE Digital, www.ge.com/digital
Inductive Automation, www.inductiveautomation.com
Malisko Engineering, www.malisko.com
Opto 22, www.opto22.com/products/product-families/groov
Rockwell Automation, www.rockwellautomation.com
Schneider Electric, www.stratus.com
“The Six Fundamental Control Strategies Every Process Control Developer Must Know;” Dale E. Reed, CAP, ISA; https://www.isa.org/WorkArea/DownloadAsset.aspx?id=153255
“Implementing a Process Control Strategy for the Food Processing Industry;” Michael McGrath, James F. O’Connor and Sinéad Cummins; The Department of Food Engineering, University College, Cork, Ireland; 1998