Comment

01 September 2003

Wireless connectivity not always a lock

Fieldbus technologies carry too many unneeded and costly options for this remote monitoring application.

By Victor Saucedo, Omar Germouni, and Ovidiu Marin

Air Liquide is the largest industrial gas manufacturer in the world.

The company provides integrated services to its numerous customers in many industries and necessarily has designed a variety of modules—cabinets—and installed them at customers' sites to ensure the optimal quantity and quality of product.

These cabinets perform similar and fundamental tasks but also have distinctive characteristics because they serve in different industries ranging from food, glass, steel, and electronics to water treatment and the like.

New information technologies can improve the economics of operating and maintaining the cabinets, as field engineers can rely on signals monitored remotely in a frequent manner to inform them when to replace a faulty part, rather than visiting the cabinet periodically even when problems have not occurred.

Remote communication also has a big impact on the operation, because this information can continuously update the plant performance. Looking at trends, engineers can diagnose the performance and take actions even before critical limits occur.

Finally, one can share the power of using this information with customers for their benefit, increasing their trust and confidence in the provided technology.

The new technologies have the general goal of transforming the signals within the cabinets into information available remotely in near to real time to any person involved in the process (while protecting it from unauthorized users).

The information generated from different cabinets, within one customer site, or by several customers should be integrated in some type of network and be available on the Internet.

Therefore, one aspect of this automation project is to investigate the suitable network that will create the desired connectivity. The other aspect is to identify the tools that will present the information on the World Wide Web.

At the same time, we studied wired and wireless networks. Numerous technologies have emerged in the past few years and will continue to do so at a fast pace in the future, and so while we have guidelines for selecting connectivity options based on specific and published needs, the problem becomes more challenging when the needs are more diverse.

Here is Air Liquide's (AL's) methodology and the steps we followed in selecting and implementing a complete technological package—software and hardware—that can be added or modified in the cabinets to improve the supervision and monitoring capabilities.

ADD CONNECTIVITY APPLICATION

These industrial modules must operate as stand-alone devices. Because the industries they serve are heterogeneous, the modules vary in complexity:

The simplest configuration consists of a group of sensors and actuators with no data communication capabilities.

Some cabinets employ AL's telemonitoring device—Teleflo. This device is capable of monitoring the modules and production through a simple modem/phone line communication architecture.

Other cabinets utilize micro–programmable logic controllers (PLCs) for local automation and control—data acquisition, actuate solenoid valves, and the like. Teleflo can also serve on these units for remote monitoring and supervision.

AL is actively searching for technologies that add connectivity to the application modules. The technologies being sought should enable the modules to communicate (send and receive, peer-to-peer communication) through an industrial network.

The information should integrate with these modules in real time and remotely through a reliable communication medium. Such information should be accessible on the Internet.

Finally, the network should remain as flexible as possible and allow new module integration/suppression with minor modifications to the existing software and hardware configurations.

The networking should also incorporate low investment and operating cost, and it should be easy to use, reliable, easy to maintain and upgrade, and have global support.

THE FASTEST SAMPLING RATE

AL's modules are five different types.

PLC type—No communication: This category applies to any module that has a set of sensors, actuators, or any other noncommunicating device.

PLC type—Communication (PLC brands): This type of module is equipped with a standard PLC that has the capability to communicate in several forms.

Teleflo type—No PLC (Teleflo): This type uses a phone line output to transmit digital and analog signals. A billing system for the delivered gases uses this information.

Teleflo type—Noncommunicating PLC: Some applications have an extra PLC (most of the time a small one) to aid in the logic operation of the module.

Teleflo type—Communicating PLC: Newer applications combine the Teleflo with a PLC that has communication capabilities.

To some extent, these modules all rely on several or more of the following specifications.

• Number and type of I/Os: A minimum and maximum number of signals per module—zero to nine.
• Real-time communications: The rate of data communication depends strongly on the network level. The fastest sampling rate is one sample per second.
• Distances: The distance between two modules, or from a module to the closest industrial computer or PLC can be short (10 meters), medium (100 meters plus), or long (500 meters plus). The ideal network should propose solutions for different distances if applicable (this implies different hardware solutions).
• Security and accuracy: Although the main objective is monitoring the modules, it is more than desirable to have an error-free, real-time communication protocol that allows one some control over the processes. Communication errors may trigger large delays, thereby preventing a proper supervision operation.
• Data acquisition and data storage: Save sampled data every second locally in an auxiliary device, as long as necessary, to troubleshoot and repair the network—from hours to days, depending on the network level.
• PLC upgrade: Recommend the replacement of the current PLC with one that has communication capabilities.

ITERATIVE NONCHRONOLOGICAL

To identify the companies that develop products with technologies that allow connectivity between AL equipment and that give AL the technological edge in customer service, the following project elements need to be clearly identified and combined:

• Available technologies: The connectivity can be wired or wireless. These two technologies are in continuous evolution and are diverse. One needs to study them closely.
• Available companies/products: Automation, electronics, and software companies that use the latest technologies release products of interest to the project.
• Compatible technologies: The selected technologies must be compatible with existing and standard solutions, as described above.

We divided the available technologies and companies/products into wireless and industrial networking. The companies releasing wireless and industrial networking products are often small companies unable to provide the worldwide support required in this project.

The selection process is iterative and nonchronological. New products, standards, and selected technologies are continuously evolving to the extent that the selected technologies may vary.

The nonchronological nature of the process accommodates the fact that several activities take place simultaneously.

A STANDARDIZED MEDIUM

Most of today's industrial networks are designed for automation and control applications. If many fieldbuses meet the specifications of the application sought, they also have many options and features that exceed the scope of the project. Remote monitoring does not need over-designed technology.

By matching the project requirements to the capabilities of Ethernet, the conclusion was that this network solution not only fits the specification, but it is also easier and the least expensive. Here's why.

Type of information: The data we are dealing with could range from raw digital or analog signals (physical layer) to serial data over classical industrial protocols (such as Modbus and Profibus).

The operators will monitor this data over an Internet-based human-machine interface (application layer). The use of any fieldbus would imply the use of several gateways and routers to match the information at different layers of the open systems interconnect (OSI) model.

Most of this equipment is expensive. To avoid these expenses, one should make the data transportable over the Ethernet medium as soon as possible.

There are many I/O modules and serial modules that are cost effective and leverage Ethernet.

Distance: Nowadays, one frequently sees Ethernet networks deployed over the factory floor. Although these networks were initially designed for office use, one can assess whether the organization can dedicate them to modular control.

In cases where there would be no network, Ethernet remains the choice because it can be easily deployed over an entire plant and comply with our distance requirements.

A larger network over which the data will be available to Air Liquide and its customers (Ethernet over fiber optics) is possible.

Network layers: Because the data collected can vary in type, the network solution should allow different layers to coexist. These sublayers will interconnect using the proper gateways to insure information continuity.

We identified several products (often referred to as e-products) on the market that can perform this task with relatively minor cost. As to Air Liquide's modular equipment, it turned out that Modbus is the privileged protocol for data communication. An Ethernet solution is possible through Modbus TCP.

Availability: The proposed industrial network shall be available in all the locations where Air Liquide operates its gas modules. This requirement disqualifies many proprietary networks, which are only available in some specific regions of the world.

Because Ethernet is a worldwide and de facto standard for monitoring applications, its availability is certain. Furthermore, its open architecture can interface with many complex proprietary networks.

Ease of use: This is one of the most important criteria. The industrial network solution for modular control should not require skills more specific than a maintenance engineer would have—either to deploy or to use.

The use of standard industrial connectors, switches, and hubs is well within the grasp of even novices.

Ease of maintenance: The advantage of the Ethernet architecture is the absence of a master/slave situation and the fact that any new item can be added to the network merely by assigning it an IP address.

The number of nodes is not constrained by any limit, except perhaps the network speed. Finally, the openness of such a network makes it easily maintainable.

Reliability: Speed is not a requirement either, and the modular control equipment shall perform perfectly even if data collects every second. Ethernet with TCP/IP is therefore reliable enough for the modular control application. The lack of determinism in the communication protocols should not affect the application performance.

Cost: The solution should be cost effective. There is no need to select expensive fieldbus technologies, particularly because we would end up not using all or most of their control functions.

Ethernet is cheaper than any other fieldbus and meets the cost requirement of modular control.

Security: The network architecture should allow for password protection with different access levels. The use of firewalls and passwords should ensure the desired security level.

Power in the line: One of the benefits of some fieldbus technologies is that power is available on the line. This feature allows cost reduction and optimizes the wiring effort. Although this is of interest in many automation applications, it does not appear essential for this monitoring application.

Retrofit: The majority of the Air Liquide gas modules are stand-alone. The solution we propose should not trigger major design changes, but on the contrary, retrofit the existing infrastructure. The monitoring equipment should also be able to become part of the future module design.

Many of today's automation products are becoming Ethernet enabled. AL's future designs should take these evolutions into account and replace the old parts (such as PLCs, HMI screens, and I/Os) with their Ethernet-enabled upgraded version.

Ethernet being the de facto industrial network for monitoring guarantees the availability of a wide variety of products. Depending on the level of communication capabilities of each category of gas modules, a set of hardware and/or software products that would achieve our connectivity needs is available.

Based on our project needs and specifications, Ethernet is the cost-effective solution. A standardized medium, it is gaining popularity with practically any device and has the robustness for supervision.

Further, when the process data threatens the information flowing over the network, a variety of devices can alleviate this problem.

PC-based devices offer the flexibility to communicate with many devices and are most acceptable when there is no control function.

NO WIRELESS GEARED FOR

The selection of a specific technology depends on the comparison of the project needs and the available technologies.

Based on the wireless criteria and commercial product research, no wireless products geared for industrial applications were cost effective for this application.

Indeed, industrial Bluetooth-based products are almost nonexistent. The most common technologies are spread-spectrum (SS) and IEEE 802.11b products. From the wireless point of view, there are only a few vendors that offer the wireless technology that meets the criteria defined in the project.

Our conclusion was that license-free technologies are more economical and attractive for modular control.

Direct-sequence spread spectrum (DSSS) and frequency-hopping spread spectrum (FHSS) have been on the market longer and have proven themselves in industrial applications. However, the type of information that they can transmit is limited, even though they meet the project requirements.

Wireless Ethernet, based on IEEE 802.11 is gaining industrial acceptance, but has less distance for rough environments. Although it is plug-and-play, one must consider other issues, such as battery, antenna, and line of sight.

In general, the implementation of wireless technologies still requires some art, and the lack of security in wireless communications keeps this technology in monitoring applications, and not in control applications. Due to the popularity of wireless technologies, organizations are using them as add-ons to either PLCs or I/Os.

FEW HAD ALL THE SOLUTIONS

The technology needed to integrate information from a variety of devices in the plant floor in real time and make it available at different levels of the control pyramid is emerging.

Having this technology and the objectives of the project in mind, an overlap is palpable. While a hardware solution will create the connectivity between different devices, software will manage and present the information to the different users.

There are many software products already using these standards. For this project to go, we needed a selection. More specifically, a software solution should meet the following requirements:

• Economical: The software solution has to cope with the cost constraints of the project. Most of the software packages are intended to work for many inputs and outputs, which adds cost to a solution that monitors few points.
• Easy to use and implement: Higher level programming languages should be used and be intuitive for implementation. Expecting staff to learn or troubleshoot software solutions in some programming language is not acceptable, because this represents a big initial and long-term investment for the company.
• Modular: The software should support a good number of devices, including major standards.
• Functionality: There are many software packages used in automation. The software solution we sought focused on monitoring.

Based on our commercial investigation, software costs range from a few hundred dollars to more than $1,000. Based on the one-day contact with the different providers, we saw that some software products are easy to implement, while others require high-level programming skills.

We also found that it is not easy to come by a software tool that enables easy implementation of data collection and display on the Internet in a cost-effective manner and that is tailored only for monitoring and a small number of I/Os.

New standards are developing that combine OPC and XML, which will enable information integration at all levels.

Finally, and as to companies that we contacted, all confirmed that they provide global support, but few had all the solutions. Advantech is one of them. The rest of the companies investigated required at least one-third party to complete the requirements.

The PC-based approach along with object-oriented programming software offered a modular easy-to-use solution for the connectivity problems.

Although the wireless solution still remains an art form, we expect it to be cheaper and easier to use in the future. IT

Behind the byline

Victor Saucedo, Omar Germouni, and Ovidiu Marin work at American Air Liquide in Ill. as process control scientist, process control engineer, and process control manager, respectively. Saucedo and Germouni are ISA members.