LyondellBasell—an $18 billion chemical company based in the
Netherlands—was having difficulty managing raw materials in plants
across the globe.
The company was facing rising prices in raw materials as well as increased costs in energy to produce its petrochemical products. The business executives needed accurate data from multiple plants in order to make effective decisions on buying and shipping the raw materials—or feedstock—to the right plant at the right time.
“We need accurate and timely feedback loops and operational models for dozens or hundreds of people working in diverse locations,” says Eric Silva, vice president of information technology at LyondellBasell. “Employees have used tools to address these challenges, but these tools did not work as harmoniously as they should.”
The company decided to integrate plant data with its enterprise systems. The project team started with the company’s North American olefins business unit, consisting of seven plants that produce a handful of petrochemicals used in packaging, detergents, clothing, car tires and coatings. The project team worked with Aspen Technology Inc., a Cambridge, Mass.-based manufacturing software supplier, using Aspen One modules to take plant data and integrate it into several applications from Microsoft Corp., the Redmond, Wash., software company.
The process for each of the plants involved taking data from Aspen One in the control room, integrating it through Microsoft BizTalk and making the data available at the front end through the Microsoft SharePoint Portal Server. The result offered cohesive plant data to help the business side with:
• Effective planning, selection and allocation of feedstock
• Translation of feedstock information into an operational plan that can be acted upon by the plant
• Online control of execution through real-time advanced control
• A rapid feedback loop based on a sophisticated modeling of the complex interactions among orders, production, inventory and distribution in the supply chain.
The task of getting critical plant data to the enterprise resource planning (ERP) system is still complicated, but interfaces and standards are making the process easier and more dependable. It’s always been done, but in the past, plant personnel grabbed the data manually. Automation has made the information more timely and accurate, but the flood of plant data can overwhelm the ERP system. Plant managers are struggling to put the data into context so the enterprise system can absorb it in a meaningful manner. This is no small undertaking. Plants produce data in real-time streams, while ERP systems like batched data.
Integration drivers
There are a number of reasons for sharing plant data with ERP systems. The overriding objective is to trim costs and make plants more efficient. This includes coordinating multiple plants to make sure the right products are produced in the right place to meet demand. Lean operation comes as a result. Data from the plant also helps companies make sure they have the right materials at the right plant. Another reason for sharing data is to meet regulatory demands, whether that’s trimming the carbon footprint or managing data for compliance. Quality is also a motive. Plant data can be used to analyze production to avoid waste.
There is nothing new about taking plant data and preparing it for the business side of the company. In the past, however, it was done slowly and inefficiently. “Moving plant data has always been done, but in the past it was done with clipboards and sneakers,” says Fred Yentz, president and chief executive officer (CEO) of ILS Technology LLC, a connectivity software supplier based in Boca Raton, Fla. “Now it’s being done in a variety of maturing ways.” Those maturing ways include electronic data exchange that delivers real-time information on production and inventory.
While companies are working to integrate the typical production data the business people need from the plant, their job is complicated by the demand for new data on everything from regulatory data to environmental information. “Plants may be challenged by a new CEO who wants carbon data to come through IT (information technology) to the enterprise system,” says Yentz. “The plant floor guy will have to figure out how to get that data to the enterprise system and do it without stopping tire production.”
The quality team is an additional customer seeking data from the plant floor. “The quality staff has been relying on plant data for a while now,” says Bob Mick, vice president of emerging technology at ARC Advisory Group Inc., based in Dedham, Mass. “If you’re making razor blades, quality doesn’t matter. But if you’re making chemicals, you need to exchange data. In process, the most important data is scheduling. In discrete, it’s inventory management.”
Real-time to batch
It’s no small challenge getting real-time plant data to the ERP system in a way that the ERP can understand and turn into useful business information. When you take data from controllers, it’s not ready for the batch processing of ERP systems. Grabbing data from controllers can also burden the controller and potentially interfere with its manufacturing responsibility. “Everybody wants data from controllers. Four systems may extract the same data at different times,” says Ron Monday, chief executive officer and president of Online Development Inc., an automation connectivity products vendor in Knoxville, Tenn. “That’s a conflict. Say they’re looking for a tank level. Four people query the controller and it chokes the controller.”
Before data can be absorbed and managed in the enterprise system, it has to be normalized for batch and business processes. “ERP systems can’t deal with real-time data. It’s not the role of the ERP to manage data coming in three-minute intervals,” says Marc Leroux, collaborative production manager at automation vendor ABB Inc., in Norwalk, Conn. “A lot of people say SAP [ERP systems] can do that, but SAP wasn’t designed to take real-time data.”
Another challenge is the high volume of data being produced by the plant. It’s far more data than the ERP needs, so it has to be filtered before it can be integrated into the enterprise system. “There’s a huge fragmentation in terms of the sheer quantity of data in plant-level systems, control, machines and all the other systems on the plant floor,” says Vivek Bapat vice president of suite solutions at SAP AG, in Walldorf, Germany. “A lot of that data is real-time or close to real-time. But the ERP is based on batch and transactional processing.”
In order to effectively take plant data and make it comprehensible to the ERP system, the integrators need to understand how plant data works as well as understand how the ERP system processes data. To get both understandings in the same room, many companies have created teams that include both plant engineers and IT personnel “It has to be people in the IT realm who understand what they need, but it’s the plant engineers who know how the data relates to the order being made,” says Keith Jones, program manager for HMI supervisory and SCADA, at Wonderware, a Lake Forest, Calif.-based automation software supplier.
Worse than useless
In order to make plant data useful, plant managers have to take the information and validate it. It also has to be put into context and sent to the ERP system through standard interfaces. If you don’t aggregate and map the data into a context for ERP consumption, the information you get can be worse than useless. “In the past, control engineers were putting plant data up to the business system and it didn’t have any context,” says ABB’s Leroux. “That’s not only meaningless, it’s dangerous.”
To make the integration of plant and enterprise effective, the plant data needs to be put into context. Otherwise, it’s like “drinking from a fire hose,” as one plant manager described it. “What we’re seeing is you have to have the information aggregated and have it in the right context, and you have to validate that it’s correct before you pull it up to the ERP system,” says Leroux. “ERP systems need to receive information that’s been summarized.”
SAP has worked to accommodate plant data for the enterprise system. The company’s purchase of Lighthammer Software Development Corp. three years ago was a move to help translate raw plant data into business process information. “We can take data from a variety of plant systems because we’re plant-floor agnostic,” says Bapat of SAP. Our intent is to serve as a filter to make the real-time plant data into information that business people can act on.”
Just as plants have moved beyond manual ways of taking plant data to the business side, they are also trying to leave behind the complicated and inflexible custom processes. Standardization is the new mantra for plant/enterprise integration. “The difference today in sending plant data to the ERP is there is more infrastructure,” says Wonderware’s Jones.
Standards such as the Instrumentation, Systems and Automation Society’s ISA95 and OPC (an open connectivity standard) have become useful tools for integrating plant data into the ERP. “The first building block in business process integration is you have to map the information from the plant floor,” says Chris Colyer, worldwide director of solutions strategy at Microsoft. “We’ve been very focused in working with ISA95 to match the business process in the ERP to the plant data that’s been mapped.”
But standards haven’t completely solved the problem of integrating multiple plants. Each plant typically has unique systems of varied ages. In most cases, each plant will involve some custom integration. “Even though there is a lot of standardization at the ERP and plant levels, what we’re seeing is that these interfaces are still custom interfaces built on a plant-by-plant basis,” says Paul Rauch, senior director at vendor Siemens IT Solutions and Services, in Norwalk, Conn.
If you build a group of new plants simultaneously, integration across the entire operation would be very straightforward. But in reality, each plant is a patchwork of old and new machines run by old and new control systems with old and new programmable logic controllers (PLCs) and networks. The complexity is multiplied when you’re working with numerous plants. Standards help convert plant data into ERP-ready information, but the process is still very complicated.
Sidebar - Integrating ERP and Manufacturing in Pharma
To read the accompanying sidebar, go to www.automationworld.com/feature-4624
