1 November 2006

Man, Machine Unite

Redundancy and human machine interface help achieve oneness in safe processes

By Frank R. Wilson

When Sir Francis Bacon said 400 years ago, "knowledge is power," he couldn't have known about automated control networks. But the phrase still rings true in more ways than one in the operation and maintenance of today's control systems. While redundancy is in great demand in the communication of control systems in today's industrial complexes, if technical support personnel do not adequately understand the overall system they are responsible for, all the redundancy in the world won't help. Yet the short fall in the advancement of programmable logic controller (PLC) reliability today is maintenance personnel do not get many opportunities to troubleshoot PLCs on a regular basis. That's where redundancy and human machine interface can be vital in successful production processes. Redundancy can bridge areas where communication of critical components may fail.

In a sawmill operating at $25 million, with machine centers each having separate PLC systems, reliable communication is of the highest priority. The loss of information about an adjacent machine center could mean catastrophe to the safety of personnel as well as to the production and damage of equipment.

Bottom line drives decisions

In today's industrial environment, there is primarily one driving force behind most decisions made in control systems. As engineers, we'd like to believe it is technology advancement. But reality tells us the real driving force is the bottom line. Control systems control a process to produce a product or service. These processes can be expensive and add to the initial cost to any industrial production operation. One of the most common means a company uses to determine whether to spend money on a project is return on investment (ROI), or the benefits of a project divided by the amount invested. But proving redundancy has a justifiable ROI is difficult. A system with a redundant communication system doesn't necessarily lead to increased production. Yet, without redundancy, you could see downtime when the system has a critical failure of the designed communication components.

When planning the system, it is important to clearly identify the scope of redundancy required to properly support the process. After purchasing and installing the equipment, the cost of adding a feature as simple as the redundant communication port may cost almost as much as the original equipment. In the case of integrators, the sales people may not communicate with the actual design engineers before agreeing to the added requirements requested by the customer. The cost of not agreeing on redundancy prior to accepting the contract could mean it is left out of the system, and the process will ultimately suffer.

Mill processes relate, duplicate

The Pacific Lumber Company in Scotia, Calif., produces over 250 million board feet of Redwood and Douglas Fir lumber per year. The sawmill is comprised of nine separate machine centers, each having an independent PLC. Platforms supported throughout the mill include Ethernet, ControlNet, DataHighway+, Serial, QNX operating system, wireless, and remote dial-up support. The mill uses these communications platforms to control input/output (I/O) and communicate ma-chine status between machine centers. The communications also handle data acquisition, human machine interface (HMI), and reporting over the company-wide intranet.

The logs come into the mill in lengths up to 55 ft. They are cut into lengths ranging from 10 ft to 20 ft, depending on the best solution from the log optimizer. They are sorted by diameter and kicked into bins to be processed by the primary breakdown sawing equipment, a double length infeed (DLI). The process that scans the logs at the DLI determines the three-dimensional shape of the log and the optimum cutting solution of lumber that we can recover. The log then rotates to the position the DLI determines. These logs process at a rate of 12 to 17 logs per minute with lineal speeds of up to 500 ft per minute. After the logs pass through the DLI, they continue on the production line and divert to one of two different edgers. Each edger has its own scanner prior to performing the next sawing phase of the lumber process. After it is cut into lumber, the log passes down a transfer chain, where graders look at the boards to give quality designations based on appearance and defects, such as decay and splits, the board optimizers can't detect. After passing through the last optimizer and being trimmed to the best length, the lumber receives an assignment location in the 70-bin sorting system and drops out of the sorter and into an automatic stacking machine.

Of the nine machine centers, a redundant main Ethernet network links all PLCs in the sawmill. Its main purpose is to pass data between PLCs and allow access for remote programming from computers located at various locations in the mill as well as dial-up access through the phone line. In addition to the main sawmill networks, each machine center is comprised of various sub-networks, depending on the requirements of the particular machine center. The I/O modules are located in operator consoles, motor control centers, and enclosures installed at strategic locations to service I/O devices such as photo cell, limit switches, and encoders.

Human machine interface key

The primary reason for process systems to fail from the control point of the process is usually related to I/O component failures: photo cell, limit switches, valves, motors, and the like. The process of identifying these failures becomes critical to getting the system back online. Using an HMI is one of the most useful means of identifying I/O problems. Most HMI systems have the trigger alarms, which can alert maintenance personnel by means of visual displays, warning lights, audio devices, and even the use of pagers. It is important to identify the critical components to design a system with the correct amount of redundancy. It's important to determine what effect a machine center or piece of equipment will have on the process if it shuts down or goes off line.

All the machine centers have an HMI connected to the communication network, except the waste system. The HMIs perform multiple functions. Besides allowing the operator to start and stop various ancillary equipment, they monitor the status of the machine center, signal warning conditions, and allow access to change operating parameters in the PLC. While the HMIs do not have redundant configurations, in the case of a failure, spare components can quickly replace the failed unit. PLC programming computers can handle the critical functions the HMI normally performs. Six remote story boards located at strategic locations throughout the sawmill cover the alarming features.

In short, the question of how much redundancy you really need in a process comes down to a few basic principles. Identify the critical operations of the system that you cannot bypass or curtail without significantly impacting production. You can accomplish this by understanding all aspects of the process. Involve production and supervisory personnel along with the design team. Make sure you give proper attention to selecting standard and reliable components in the planning and installation process. Assure the system is fully supported with spare parts in addition to needed redundancy, and make sure you communicate the process to everyone involved in the operation.


Frank R. Wilson is electrical/automation manager at the Pacific Lumber Company in Scotia, Calif.

Fast Forward

  • When humans and automation interact, communication and redundancy are critical.
  • While the bottom line is cost efficiency, redundancy still rules the roost.
  • A California sawmill uses redundancy and HMI to ensure processes run smoothly.

Human reactions in three-dimension

By J. Kang, K.K. Wright, S.F. Qin, and Y. Zhao

Human reaction in hazardous situations can make the difference between life and death, especially in manufacturing scenarios, where humans and machines must coexist on a daily basis. Plenty of research exists on human modeling and simulation. Several research groups have worked on virtual humans in the computer games industry, for instance, but we found no reported literature on the real-time simulation of human reactions in a hazardous environment. This type of research could be instrumental in developing systems where effective human interaction is essential.

Yet, since human beings cannot put themselves in a real-life dangerous environment and run tests, we developed real-time simulation of human reactions in hazardous environments. In order to make the modeling and simulation as realistic as possible, we employed the latest 3-D computer animation technologies to develop a real-time simulation system. We classified conscious and subconscious behaviors and reactions for groups of people encountering hazardous situations, and how the emotion and personality and anxiety and fear affect their decision-making. We constructed character body 3-D surface models with a 3-D scanner and captured motion postures such as walking, running, jumping, falling, climbing, crawling, and being afraid and astonished.

We established 3-D models for the scene. While some software programs already have rooms and furniture, we needed to create dangerous situations, such as fire burning, smoke, explosions, and collapse.

Future research

Currently, human motion modeling can come to life through motion capture system and 3-D computer animation software, whereas to integrate perception and intelligence into virtual humans' motion is still a huge undertaking. Artificial intelligence, expert systems, and robotics have been researched for a few decades, and many of these techniques could model virtual humans' perceptions and intelligence. One challenge is the synchronization of motion and intelligence.

Enabling perception, intelligence, feelings, and emotions of a vivid virtual human interactive with a virtual world on the computer screen is a new research area. Another big challenge is to precisely model virtual humans' perception. This is just a beginning. We are now evaluating the techniques in restricted environments.

About the Authors

J. Kang, D.K. Wright, S.F. Qin, and Y. Zhao are students at the School of Engineering and Design, Brunel University, Uxbridge, Middlesex, U.K.