Enable floor to enterprise integration

Think Again: Automation and controls increasingly are being integrated with enterprise-level applications through the manufacturing execution systems (MES) and manufacturing operation management (MOM) systems.


Plant to enterprise integration and other issues related to Control Engineering and Plant Engineering were among topics discussed when Dennis Brandl, chairman, MESA Americas board of directors, and Michael Yost, president, MESA International, spoke with CFE Media’s Steve Rourke and Mark T. Hoske about system integration and manufacturing efficiency.

Probably 70-80% of companies still could benefit from more effective use of enterprise-level applications through the manufacturing execution systems (MES) and manufacturing operation management (MOM) systems, Brandl said. Those not effectively using MES/MOM may be using spreadsheets or customized connections between the floor and enterprise systems. Using standards-based architectures and best practices is key to efficient and effective integration to get the best corporate value, he suggested.

Brandl said that perhaps 70% of companies could be applying tools that use ISA 88 (batch control, also good for broader modular software applications) and ISA 95 (plant to enterprise) standards. Many know they have a problem, but don’t know that a solution is available. Many hear that they need lower cost solutions for analysis of “big data” analysis, though perhaps not necessarily enterprise resource planning (ERP) software. On that topic, “I think we’ll see some movement within the next 9 months,” Brandl said.

Yost expects to see a greater push toward understanding the return on investment (ROI) for MES and MOMs. At present, most connectivity from the plant floor to the enterprise and beyond is very narrow, addressing a particular problem or perhaps a broader issue by application. This can lead to the next step, increasing the scope of application.

Brandl said another approach has been more of a strategic directive rather than tactical problem solving. Bigger companies have productivity issues related to spending $500 million or more on ERP systems when the data needed to make ERP valuable just isn’t there.

Solutions providers are adding it all up and helping customers to get that value back, which can be substantial. Brandl said examples include Cargill, Merck and Johnson & Johnson. Often, the solution isn’t installing 50 ERP terminals, but using MES effectively to put data where it needs to be. The ISA 95 effort focused on the integration of data from programmable logic controllers (PLCs) and distributed control systems (DCSs) with business level systems, Brandl said, with MES serving as the magic in between.

“Initially the problem was that we needed to integrate plant floor and business systems to fully understand the whole value we were getting. Value comes in automating workflow, reducing error rates by half, cutting rework, while increasing quality and customer satisfaction,” Brandl said.

Some buy MES because they put in an ERP system. Only part of custom systems can be replaced by ERP software, Brandl said. ERP systems aren’t designed for things like obsolete inventory, weighing, and dispensing. Others have very specific issues, such as trying to track equipment for regulatory reasons. Regulatory compliance can be the biggest push, once ERP is installed.

MES makes ERP productive

MOMS and MES are ERP’s interface to the real world, Brandl said. Consider it a 10%-15% tax on the cost of the ERP system to make it more effective and get past the lament: “They told me we could do everything with this.” The ISA 95 standard applies the MESA models, to plant, lab, maintenance, and other manufacturing operations, providing site level details. Laboratory information management systems (LIMS), tank farms, automated warehouse systems, and broader operations all can be integrated components.

Brandl said that in recent MESA meetings in Copenhagen, those gathered discussed the changing role of information technology (IT), cloud, big data, and bring your own device (BYOD), and what these trends mean for manufacturing.

Cloud connections

The cloud will continue to transform computing into a utility, just like water, power, or an electric or gas utility. Whether local or public, it will just be there and have the uptimes associated with current utility services. Smaller companies may avoid buying servers entirely. Others may need local services to decrease risk of disconnection. The cloud changes how companies specify, order, and deal with software needed at this level, Brandl suggested, and in the future, IT will be split into groups that understand software and those that understand servers and networking.

While there’s a lot of talk now about big data, process and manufacturing facilities have always had “big data” issues, Brandl said. Dow Chemical, for instance, produces 23 gigabytes of data daily, and that’s compressed. Big data without context, however, cannot provide information. Workflows and recipes of MES provide context that helps transform data into information. Software provides analysis to get information from data via automation, historian, workflows, recipes, and batches.  

Yost said some companies are collecting all the data they can, because memory is relatively inexpensive, but without structure, it’s just random data. One machine tool manufacturer collected 13,000 downtime reason codes, and Gigabytes of data crashed the system. Those involved pared that down to the six most critical error codes they should collect.

<< First < Previous 1 2 Next > Last >>

Jonas , , 11/20/13 04:19 AM:

I agree the ERP need raw data input in order to generate an information output. Since information is distilled from a larger set of raw data, it gives the plant must collect all the raw data needed to obtain the information.

Indeed the raw data originates from below the DCS and the PLC. The data comes from sensors located at “floor” level 1 of the ISA95 functional hierarchy model. Traditionally sensors use hardwired 4-20 mA and on/off signals, but increasingly one of many “H1 fieldbus” technologies are taking the place of hardwired signals, in my personal opinion, to more efficiently cope with the larger number of sensors, and the larger number of signals from each intelligent device. The particular “H1 fieldbus” protocol used depends on the industry since each industry has different requirements. The process industries tend to use FOUNDATION fieldbus H1 or possibly PROFIBUS-PA while discrete manufacturing may use IO-link, ASI, or CompoNet. The wireless sensors use WirelessHART. The “H1” fieldbuses used by sensors and actuators at level 1 should not be confused with the “H2 fieldbus” (such as Modbus/RTU, PROFIBUS-DP, and DeviceNet) or increasingly the industrial Ethernet (Modbus/TCP, PROFINET, or EtherNet/IP) used at level 1-1/2.

When existing plants were built maybe 20-30 years ago, they were built with a bare minimum of instrumentation (on the P&ID) to operate the units because 4-20 mA wiring and system I/O cards are expensive. However, plants are now pushed to increase uptime, reduce maintenance cost, and become more energy efficient etc. This requires more automation which means more sensors. Running more wires carries the risk of damaging the existing plant if cable trays and junction boxes are opened. WirelessHART is a good way to modernize existing plants by instead deploying wireless sensors throughout the plant to cover these “missing measurements” as a second layer of automation beyond the P&ID. The data often goes beyond the control room; to maintenance and reliability office, into software that takes raw wireless data from vibration and temperature transmitters into asset monitoring software where multi-parametric algorithms extract equipment health information in real time helping the plant to ensure the equipment operates efficiently and to plan maintenance for the right time. Raw data from wireless acoustic transmitters and flowmeters goes into energy management applications to extract steam consumption information on a per plant unit basis for cost accounting and steam trap monitoring software extracting steam system health information driving replacement of faulty steam traps. This is pervasive sensing:

Personally I also see a need to transform raw data into information in real time in addition to storing data in the historian for later analysis. Real time information can be acted upon at once, without delay. I see level 3 software as important to reduce the entropy of information; converting raw data into actionable information:
click me