Pouring Thought into the Process

Software for control

Embedded intelligence

Networks & communication

Process control & instrumentation

Machine control

Human-machine interface

Sidebars: Distributing Intelligence across the Information Architecture Fired up, at all hours, with distributed intelligence Distributed intelligence polishes machine design ONLINE

Designing to incorporate distributed intelligence has advantages, but can include challenges outside the realm of traditional thinking. The world over, thousands of automation and controls manufacturers have poured a lot of thought into where intelligence should reside in automation and control system design. In the interest of reducing costs and making automation and control more effective, distributed intelligence has swelled from a drop in the bucket to a sea of change.

Ability to distribute intelligence impacts multiple areas of information architectures. See a sampling of influences, starting on the facing page.

Distributed intelligence allows a “sensor, actuator, human-machine interface, or other node to independently process local or network-generated information; make a decision about how to react to data; and be capable of sharing both the information and decision with other nodes,” according to Michael R. Tennefoss, a director at Echelon (Palo Alto, Calif.). Echelon offers products and services related to LonWorks networks–which spread widely across factory, utility, home, building, and transportation markets.

S. Zafar Kamal, global business manager, Plant Optimization Products, ABB Automation (Rochester, N.Y.), says distributed intelligence supports decisions humans make by “eliminating mundane/repetitive tasks, providing data filtering, data organization, data crunching, or generating feasible/optimal alternatives or scenarios, so that the person is able to concentrate on making the right decision.”

Automation, Dr. Kamal continues, “enables human resources to focus more on decision-making and less on data gathering. Subsequently, the end-user should expect intelligent systems to manage and filter alarms, highlight relevant exceptions, predict failures, predict quality, take over mundane tasks, help optimize capacity, and generate optimal operating conditions.”

Enabling multiuser software gives operators more power to share information across the enterprise, suggests Craig Thorsland, Cimplicity product manager, GE Fanuc Automation (Charlottesville, Va.). “Today, the operator still plays a key role in interpreting information that is gathered and presented. As technology advances, the level of automatic decision control is advancing as well. This will allow the empowered workforce to take on higher and higher levels of responsibility as their knowledge and tool set continue to expand.”

Gateways, field intelligence

Applying technologies of distributed intelligence architectures in field devices in various locations lowers costs and increases performance, says Camille Crouch, technical specialist, Siemens Energy & Automation Inc. (Alpharetta, Ga.). Kinds of devices include the following:

Multiple network gateways often are embedded systems that link existing network architectures. This communication hub receives data from multiple field networks, coordinates and exchanges data among different field networks, and transmits the data to higher level systems via one established network, such as Ethernet;

Data concentrator host devices, similar to gateways, collect data from networked field devices to perform data management and storage. Stored information is periodically provided to a host and gives the ability to hot-swap devices, readdress nodes, and other tasks to maintain system robustness;

Soft logic control on a PC can effectively use smart field devices, such as intelligent valves and sensors; and

Embedded control intelligence at the field-device level improves with decreasing costs for sophisticated memory chips (flash technology) and shrinking processor chips (molecular chips).

Pushing decisions into devices delivers other advantages. Echelon’s Mr. Tennefoss says, “Distributed processing allows nodes to process inputs and outputs and complex algorithms without burdening the network with extra traffic, as well as operate in the event of a network failure. The network provides a means of communicating among nodes, downloading new applications into the intelligent devices, uploading error and history statistics, and remotely viewing system activity via telephone or IP routers.” Distributing intelligence increases availability by eliminating single failure point locations, which simplifies maintenance and upgrades since information is available in multiple locations, and extends connections outside the plant, he adds.

Examples: I/O devices, networking

Network standards and continued cost reduction of networking have played a great role in spreading distributed intelligence, says Stan York, FlexLogix product manager, Rockwell Automation (Cleveland, O.). “However, just as significant is the developing standardization among control platforms, which simplifies dealing with various control disciplines and control ‘boxes,’ because users will have only one programming environment and one human-machine interface [HMI]. This greatly reduces the learning curve and makes set-up and configuration simple,” he says. One implementation is when vendors move toward one programming environment based on one execution engine that can be leveraged in distributed, motion, process, or drive applications.

An embedded networking chip manufacturer offers several application examples. William Peisel, vp of engineering and chief technology officer, NETsilicon Inc. (Waltham, Mass.), says, “Today’s variable-speed motors have almost no visibility from the HMI level; therefore, adding new control functions are difficult or impossible. With distributed intelligence-enabled via a microprocessor and controlled via HTTP over Ethernet-that problem goes away. Another example would be adding intelligence at the I/O level, which would be a simple and low-cost way to improve access to end points.”

Charles Piper, Foxboro (Foxboro, Mass.) I/A Series system product line manager, agrees, saying, “I/O and servo devices are excellent examples. Using intelligent remote I/O solutions, it is possible that the savings in purchase costs of wire alone can be larger than the purchase cost of the I/O products. Economic benefits of distributing intelligence are greatest when the device will need a fair amount of CPU horsepower to perform its basic functions. In this case, adding additional code for embedded automation intelligence-such as control functions, servo functions, or increased diagnostic capabilities-results in a negligible cost increase.”

Logic and intelligence will move from centralized to mixed to distributed based on customer needs, says Brian Brickhouse, Cutler-Hammer’s (Westerville, O.) automation product line manager. Present locations for field device intelligence include motor starters, sensors, IO modules, drives, and user-interface products. Now, an intelligent motor starter can monitor and communicate phase currents, control voltage, and ground fault and phase unbalance conditions, he says. Coming soon: more predictive diagnostic information and ability to execute logic in field devices.

The control loop-sense, decide, actuate-remains constant, but the tidal wave of distributed intelligence disrupts products’ form and function to cost-effectively meet user’s needs.

Distributing Intelligence across the Information Architecture

Intelligence-ability to make decisions based on gathered data-can be distributed closer to and across, the process. Changing technologies and trends influence information architectures and ways of thinking about control and automation system design.

Decision Location

New: Decisions occur more easily in multiple locations, within a central control system and various workstations, as well as embedded in individual machines, distributed hubs or I/O blocks, devices, and even in sensors. System design focuses more on function and less on product type. Products combine and blur in definition and functions. Software goes almost everywhere.

Traditional: Decisions occur in fewer locations, such as a programmable logic controller, loop or batch controller, or a distributed control system. Upgrades often are product-focused; hardware is more prominent.

Data Flow

New: Ability to use a few or one digital network type throughout the enterprise reduces network hardware costs. Network software may vary, but information flows more readily. Some plant-level networks remain separate to ensure reliability.

Traditional: Digital networks, if used at all, are proprietary, requiring dedicated hardware and software. Data transfer among networks requires more equipment and support and may not be in the most usable form.

Higher-Level Communications

New: Information about any part of the process is available from multiple points within and beyond the plant system. Trend is toward exchanging information between plant and enterprise systems in both directions through multiple locations.

Traditional: Within the plant, data feeding decisions remain isolated or unavailable. If an electronic connection exists from plant to enterprise systems, data often flow only up through a single point.

Programming Environments

New: Simplified programming occurs in a single environment and can be scalable, distributed to multiple layers or locations with little extra effort. Programmers for more readily available commercial languages need to understand needs of industrial settings.

Traditional: Multiple programming environments can cost more to support and modify. Training also may be more of an issue. Information in multiple forms requires updates in multiple locations.

Silicon and Software

New: More economical silicon and software based on commercial solutions help distribute intelligence within and across platforms. Industrial computing incorporates ever-faster commercial chips and boards in multiple form factors and locations. More frequent upgrades aim to preserve existing investments, but may not need to if large returns more than justify installing new technology every five years or less. Code reuse and lifecycle costs are considered.

Traditional: Traditional: Proprietary silicon and software, often great for longer-term point solutions, is usually less economical to reconfigure, maintain, and adapt to other vendors’ devices. Even though design life may be 15 years or more, buying decisions might only look at capital costs.

Vendors

New: Limited user personnel rely on system integrators or “families” of vendors that offer support for using best-of-breed, compatible solutions incorporating distributed intelligence, which may need adjustments to configure and operate as expected.

Traditional: Buying from a single vendor provides a single point for contact. More user alternatives have helped advance distributed intelligence technologies, even for users choosing to buy from one or very few vendors.

Standards

New: Users and vendors move toward market-based standards for price advantages, but still rely on groups and organizations to set guidelines to encourage interoperability, creating options for moving the point of decision to where it makes the most sense.

Traditional: Committee-based efforts create stability, but limit changes and perhaps innovation in distributed architecture decisions. E-mail has helped reduce time.

Optimization/Maintenance

New: Intelligence across the system means various points check in prior to failure, so maintenance can be scheduled and parts delivered in time. Troubleshooting and optimization can occur even from outside the plant through web browsers, drilling down to device level, with little or minimum disruption.

Traditional: Sometime the line or process stops for hours, unplanned, until the problem is located, diagnosed, and corrected.

Source: Control Engineering

Fired up, at all hours, with distributed intelligence

Stanford University (Stanford, Calif.) relies on “distributed intelligence” to deliver steam and electricity for mission-critical studies at all hours.

With a service area including the university’s medical facility, Cardinal Cogen, also in Stanford, owns and operates the on-site power plant, a 49-MW, natural gas-fired turbine generator, providing steam and electricity.

“Right now, power reliability at Stanford and its medical center is our foremost concern,” says Jim Soutter, operations manager for Cardinal Cogen.

“However, our ability to be a cost-effective producer in the utility market grows more important every day in light of deregulation. That puts pressure on controlling operations and maintenance costs and improving system performance. This system with distributed intelligence allows us to meet both objectives,” Mr. Soutter says.

GE Fanuc Automation (Charlottesville, Va.) products contribute to the solution. The plant uses GE Integrated Control Systems (ICS), Cimplicity HMI software, Series 90-70 PLCs, Genius I/O, and Field Control I/O.

Genius bus controllers in the PLC cardracks communicate remotely with the appropriate Genius or Field Control I/O block, which is wired directly to a sensor or actuator. Each block has its own microprocessor for communications, monitoring, and control. For example, one Genius block has the specific I/O function of reading a thermocouple input. Likewise, input can come from up to eight Field Control I/O modules in a vertical rack called a Field Control Station. While the I/O block communicates with PLCs over a Genius bus, servers receive data over an Ethernet LAN.

Monitoring and control stations throughout the facility give plant operators easy access for fast response. “We can monitor and adjust controls from several convenient locations without having to go to each individual block,” Mr. Soutter adds.

Distributed intelligence polishes machine design

Incorporating distributed control allows a leader in belt grinding and polishing machines to simplify and centralize arrangement of operator stations.

Löser (pronounced LO-zer) GmbH (Speyer, Germany) manufacturers machines for processing metal, ceramics, plastics, glass, wood, and paper.

The system pictured consists of four grinding stations and two polishing stations for processing steel rods before and after chrome plating. After chrome plating, another three grinding stations and four polishing stations follow-with an additional superfinishing unit with two heads positioned between these stations.

Distributed I/O design–using Profibus, Siemens Simatic ET 200M and on-site switching cabinets–provides flexibility to consolidate separate operator panels into a central Microsoft Windows CE-based Simatic MP270 multipanel from Siemens (Alpharetta, Ga.; München, Germany), which controls and monitors all stations.

Safety functions such as emergency-off retain conventional design. Following customer requests, Löser also replaced the previous separate digital displays for current consumption of the grinding motors with bars on the MP270-so worn out grinding belts can be detected faster.

ONLINE

Find out how systems on a chip help shape distributed intelligence, with comments from the technical director of a group of industrial computer board manufacturers.

Do you have experience and expertise with the topics mentioned in this content? You should consider contributing to our CFE Media editorial team and getting the recognition you and your company deserve. Click here to start this process.