Systems Integration

Pathways from operations to enterprises are getting shorter, simpler

Avenues and hurdles are getting streamlined thanks to more capable components, software and networks, but the primary task remains convincing potential users.

By Jim Montague

It used to be a long, twisting trip from operations on the plant floor or out in the field back to the administrative and enterprise levels. All kinds of added devices, programming, signal conversion, networking and other checkpoints were navigated to get data from where it was generated to where decisions could be made, and these snags stifled many efforts to integrate plant-floor and enterprise levels.

Well, many of those avenues and hurdles are getting shorter thanks to more capable components, software and networks. In fact, the tools have achieved so many gains recently that the biggest obstacles may be in the minds of potential users, who mistakenly believe the production-to-enterprise trip remains longer and more difficult than it really is.

For instance, when specialty vegetable fats manufacturer AarhusKarlshamn Sweden AB migrated from manual reporting to 800xA process control with Smart Clients software from ABB, it also tied in its business system. Together, they automatically feed production orders into AAK's hydration plant, and report on which and how much of each material is used. The hydration plant is where the fats’ melting points are analyzed and determined according to each intended application, such as substituting for butterfat or cocoa butter (Figure 1).

"Integrating our control and business systems more closely created an unbroken data chain from customer order through production planning and process control," says Anders Petersson, lead automation engineer at AAK. "This increased productivity allows more secure production, provides valuable feedback that optimizes our raw material purchasing, and simplifies life for our operators."

In addition, Smart Clients lets AAK collect different types of production data in its office, where staff can follow up on key values such as energy consumption, or use asset monitoring devices to signal when equipment service is needed. "We're also using batch control with specific parameters for each batch," adds Daniel Knutsson, automation engineer at AAK. "Each batch is fully traceable, so we can see exactly when it was processed. It's also much easier for our developers to change an existing recipe, or create a new one without disrupting production."

In another move from manual, Debswana Diamond Mining Co., Ltd. in Botswana needed to replace its old, standalone DAS server and emailed reports at its Orapa, Letlhakane and Damtshaa mine's (OLDM) Plant No. 1 with a more dynamic data mining and reporting system. This solution needed to collect real-time, ISA95-compliant data about production and performance, and deliver it to all areas of Debswana's business to improve optimization and resource allocation.

To that end, Debswana recruited South Africa-based Bytes System Integration to implement an ArchestrA system platform at OLDM, including a Galaxy data repository, application object server (AOS), and Wonderware Information Server (WIS) and Historian from Schneider Electric. Plant equipment data is published for analysis in nine WIS reports, and a daily dashboard outlines key information, such as carats and ore tonnages on one screen for all WIS users on Debswana's intranet network.

“One of the most compelling aspects of the Plant No. 1 project was moving from manual information capture to automated data retrieval," says Zwikamu Dubani, IT analyst at Debswana. "This not only greatly reduces errors, but also speeds up the delivery of accurate information. Now, I no longer have to worry about ‘death by spreadsheet.’ ”

Fewer steps saves steam

Likewise, Denka Singapore Ltd.'s polystyrene resins plant on Singapore's Jurong Island buys steam from a local utility, and operates hundreds of steam traps that can fail over time, wasting steam, causing erosion/corrosion and reducing heat-transfer efficiency. The chemical company usually does periodic surveys, and hires contractors to inspect the traps annually, which means accepting some steam loss between inspections. However, it recently added Rosemount 708 wireless acoustic transmitters to 149 of its critical traps, and also began using monthly, subscription-based Remote Monitoring Service from Emerson Automation Solutions.

The transmitters monitor noise and temperature of the steam traps in real time, and transfer data via edge gateways and a wireless 3G network to a Microsoft Azure virtual cloud server. SteamLogic analytic software analyzes the data and generates alerts; experts at Emerson's PlantWeb Center of Excellence review them and report back to Denka; and Denka’s maintenance staff repairs or replaces failed steam traps according to the reports, using new standard operating procedures (SOPs), which saves steam and makes periodic and annual inspections unnecessary.

“The exception reports provided by Emerson become work orders for the maintenance team, and enabled us to reduce steam consumption by 7%,” says Ng Hock Cheong, maintenance manager at Denka Singapore. Besides saving on steam, remote and continuous monitoring at Denka reduces its traps in bypass mode to less than 4%, and identified 15% blow-through and 8% cold units on startup.

"Users want to connect to the field in different ways, and that means expanding beyond the usual DCS to enterprise areas, especially for measurements that aren't core to control, but can help plant performance and reliability," says Moazzam Shamsi, director of global solutions architects at Emerson. "The major oil and pharmaceutical manufacturers are working to understand how to move data into broader architectures. Not everything needs to go through the control system, so they're routing data around the edge of their DCS to new and existing applications."

Shamsi adds that PlantWeb Insight runs on Layers 3 and 4 of the seven-layer Purdue control hierarchy model, which is outside of Layer 2 where its distributed control system (DCS) operates, though they're all part of the PlantWeb Digital Ecosystem. "This also makes adjustments easier because, if a user wants to add measurement points for a pump on a heat exchanger, they'd traditionally have to add I/O they may not have space for, and go to the vendor to add algorithms to the DCS for those new functions. Picking up applications outside the DCS avoids this because, while the DCS keeps plant performance on target, outside data can help manage the business without being directly attributable to the DCS or its requirements. Also, where users previously had to buy a whole DCS infrastructure, they can now pay per-tag for service and scale up as needed, which changes the whole process control business model."

{pb}

Cooperative multitasking

Similarly, Dr. Reddy's Laboratories Ltd., Hyderabad, India, began its latest quest for stable production volumes, compliance and performance optimization through better data in 2010 at its active pharmaceutical ingredients (API) plant in Visakhapatnam, India, where it's driving operational excellence by expanding its combined DCS/manufacturing execution system (DCS/MES). The project's first phase included 9,800 I/O points, servers and software that began operating in 2014, and a scale-up of another 10,500 I/O and support components that went live in December 2016. Equipment monitored and managed at the plant includes reactors, centrifuges, dryers, weigh scales, barcode scanners and other support devices (Figure 2).

"We want electronic logbooks linked to our historians, recipe workflow execution, and batch control data because we need manufacturing intelligence," says Girish Deshmukh, vice president of engineering and projects, Dr. Reddy's. "Then we can add overall equipment effectiveness (OEE) and other data sources, and generate batch reports and verifications about production downtime and moving quality targets. We began working with Rockwell Automation when we integrated the DCS and MES, and enhanced operational transparency to improve quality and meet regulatory norms."

Together, the two firms integrated and connected the plant's quality by design (QBD) and supervisory control and data acquisition (SCADA) functions like online recipe management, and implemented an action plan to enable multivariate analysis and other capabilities. "As a result, operators didn't need to switch screens as much anymore because they could view both DCS and MES displays at the same time," adds Deshmukh.

Because the API plant used to have so many manual operations and documentation, Deshmuhk reports all these human interventions could slow it down. "What we really wanted was to get to a paperless plant, where the SAP enterprise system could send an order to the MES, which could examine available stock and tell the DCS and batching systems to pull ingredients and begin processing them," he explains. "We wanted to standardize on individual recipes and have one version of the truth, so production could be managed and maintained by regular operators without knowing anything about software programming."

Consequently, the plant's architecture was revamped to include PlantPAx as its DCS, PharmaSuite at the MES level, and SAP for enterprise resource planning. These enabled common views into operations, while also reducing software and spare equipment inventories, training and control hardware.

Deshmuhk adds that Dr. Reddy's learned several valuable lessons from implementing its combined DCS/MES architecture at the API plant. "We found out about incompleteness of business requirements, the underestimation of workloads and resource availability, and how much learning and adaptation the new system would need," he added. "We also learned it's important to incorporate site-specific feedback during deployment, organize change management, and have adequate resources with each partner. However, the result is we now have one batch ID for accessing everything, and the system captures all the data. The MES shows us all deviations on dashboards. This gives us quick changeovers when we need to manufacture drugs fast, and the flexibility to design new recipes and products when needed. We want to put all our plants on a combined MES/DCS platform."

Jason Wright, PlantPAx business manager at Rockwell Automation, adds its Connected Enterprise program is gaining acceptance among existing business structures because it can access plant knowledge, make users more competitive, and balance openness with security. Its tools include FactoryTalk Cloud networking, FactoryTalk Analytics software and Connected Services. "This is all about driving data access to the best place for making decisions," says Wright. "We're really at another new inflection point. Just as operations technology [OT] and information technology [IT] are coming together, we're also seeing controls and business systems integrate, merge and become more alike. Instead of the costly, custom interfaces of the past, we're moving to the flatter, converged, plantwide Ethernet [CPwE] we've talked about with Cisco.

"For instance, a medium-voltage drive with an Ethernet port has a lot of intelligence. Previously, we had to go through the whole control system, but with CPwE, users can choose the path they want. They can drive data to the right layer for the best decisions, such as sending it to Connected Services for maintenance or where it's needed by the enterprise.”

Software + Internet = Cloudy forecast

Of course, as the path from plant to enterprise shortens, it's also enabled by more software, Internet links and cloud data storage and analysis."The process industries are moving to distributed intelligence, putting algorithms on low-cost hardware, and using wireless to bridge barriers, so fewer rigorous implementations are needed to reach safety areas," adds Michael Harmse, senior director of asset performance management at Aspen Technology Inc. "Low cost means these solutions can be added immediately, and send information to a data lake or the enterprise level without shutting down applications.

White Paper: Essentials of Big Data, Distributed 

Harmse report there are numerous unmanned air-separation units providing oxygen to their plants worldwide, and where they usually save data to their historians, they're now sending diagnostics to the cloud and their operators and managers via business systems. As a result, users are adopting enterprise historians for data analysis, which also allows them to more easily compare performance at multiple sites.

"In fact, AspenTech's Information Plus (IP) 21 used to be a regular historian, but now it's a real-time enterprise historian, which is a great data source for predictive maintenance and analytics," says Harmse. "The world's data is being pulled into huge analytical systems with multiple, converged technologies, which use empirical data models, machine learning methods and even artificial intelligence. These can quickly provide detailed calculations, which used to require specialists and take a long time.

"Now, users can employ 'genius in the dashboard' functions online, and see how a plant is running. This means users no longer have to say, 'If we'd only known 30 minutes sooner,' and they can avoid mistakes and see much earlier when equipment is expected to degrade. Even very small companies can quickly pump their data to a third-party cloud such as Amazon Web Services, and AspenTech can deploy VMware to manage it on their behalf."