01 October 2002

Building the right network

Realistically, few plants rely on a single fieldbus technology . . . and why should they?

By Ian Verhappen and Eric Byres

When a major food company began considering the use of new fieldbus technologies, it realized these technologies were a significant departure from the standard control strategy.

The company developed an economic model to determine the best possible combination of these buses. This model is unique in that it considers a variety of buses simultaneously.

In addition, the model factors in cost items such as the I/O mix in the plant, the wiring layouts, and engineering costs in addition to the usual materials requirements.

The bottom line is that the model showed fieldbus usage saves about 30% in project costs. Further, the model said the ideal implementation involves two or more different buses in a facility rather than a single bus.

Subsequent checks of the model's results with user interviews and literature searches to compare the cost, maintenance, and engineering issues of the various fieldbus solutions added further weight and urgency to adopt a fieldbus, regardless of vendor.

Use this model to select the right bus.

DIGITAL VEHICLES

There are at least 27 different bus technologies currently proposed as possible industrial field or device buses. Because this is an unwieldy number to properly analyze in depth, we chose four based on current market share and the bus being a true fieldbus-which is to say a digital, two-way, communications link for intelligent field devices. These criteria produced for the model AS-i, DeviceNet, Foundation Fieldbus H1, and Profibus-DP.

AS-i is the simplest and least expensive of all the buses. It allows discrete devices as small as push buttons or photocells to connect via bus. The data payload is only 4 bits, so it isn't useful for analog control. It is very effective for simple devices. Typically, AS-i serves the discrete manufacturing industries or other applications with high discrete I/O counts.

The physical layer technology is a special flat cable or standard shielded twisted-pair cable. Both allow field devices to operate over the bus. The topology is trunk and drop, and distances are strictly limited to 100 meters. AS-i uses a memory location mapping scheme and has no true device descriptor system. It is a single master bus and does not provide redundancy.

Its major strength is its absolute simplicity.

DeviceNet is based on the controller area network, or CAN, protocol and is targeted to the manufacturing industries, with the majority of its products used for multiple discrete I/O and array or digital signal devices such as motor starters, variable speed drives, and other on/off devices.

The physical layer technology is a special five-conductor cable that allows field devices to power over the bus. The topology is trunk and drop, and distances are strictly limited to either a 100-meter or 500-meter trunk, depending on the cable used. A DeviceNet segment can have a maximum of 64 devices.

DeviceNet's media access control is carrier sense multiple access/collision avoidance, but it usually runs in master/slave mode. This provides the bus with good time synchronization but with no possibility for redundancy.

The devices are defined to the host systems through the use of an electronic data sheet, a standardized ASCII file that provides a description of the device, including a definition of the data available for transmission.

The Profibus family contains three different buses: Profibus-DP (distributed periphery), Profibus-PA (process automation), and Profinet. Profibus-DP works in this model because of its dominant market share.

Profibus-DP is a discrete device bus using master/slave architecture. At the physical layer, Profibus-DP uses EIA-485 and asynchronous technology. The data rates run from 9,600 bits per second to 12 megabits per second, but the high speeds come at the expense of very short run lengths-100 meters-and a fairly restrictive daisy-chain topology.

As a result, Profibus-DP rarely runs at data rates above 500 kilobits per second. Field devices cannot take their operating power from the bus, and when adding new devices, the bus must shut down.

Profibus-DP uses a memory location-mapping scheme, compared with a tag-based system such as DeviceNet or Foundation fieldbus. Vendor-prepared device database (GSD/ GSE) files provide the system with device characteristic information. The GSD file is the device data sheet file, while the GSE file is the device database file.

Profibus-DP targets industries and devices similar to DeviceNet's, predominantly motor starters and other discrete signal devices.

Foundation fieldbus (FF) comes in two flavors, H1 and high-speed Ethernet (HSE), though at present there is a limited number of HSE devices or applications. As a result, this model focuses on the H1 protocol.

The H1 protocol operates at 31.25 kilobits per second, is loop powered, and supports segments up to 1,900 meters total length. A vendor-prepared device description (DD) file describes the functionality of a FF device.

The DD file also requires an associated common file format ASCII text file that a host system can use to configure the system off-line. Until recently, FF has had limited support for discrete applications, though a number of larger discrete valves and field-mounted discrete I/O stations are now available.

FF also supports redundancy of both protocols and uses cyclic communications to publish and subscribe to data on the network.

Devices can be added and removed from the network while it is live because the devices are "recognized" by the host when they are plugged in or removed. FF is the only fieldbus protocol that supports control in the field.

Fieldbus figures in the continuous process industries, with its strength in analog control of single and multivariable field elements.

DETAILING COST SUMMARY

This economic model compares the four buses, along with traditional hard-wired installation costs. This model differs from most because it allows the user to assign a mixture of devices to different fieldbuses.

For example, from a list of devices, it is possible to assign the analog transmitters to FF, the drives to Profibus-DP, and the discrete valves to AS-i. This represents a more realistic representation of the typical fieldbus installation today, as few plants use only a single protocol.

At the core of the model are three different types of spreadsheets for data inputs and calculations:

The constants sheet handles costs and rates associated with equipment and labor.

The bus details sheet contains the main calculation area and detailed summary of the costs associated with each bus type.

The inputs and results summary is a table to compare various combinations of bus installations and then see the impact of those changes on the total installed cost.

The most important worksheet in the model is the one containing constant costs. It forms the basis for all subsequent calculations. The worksheet itself is divided into five tables:

Design criteria is where plant design preferences such as connectors, conduit vs. tray, and average cable trunk or drop lengths list.

The device cost/availability table is where one lists each bus protocol's device price. This information serves to calculate the difference between the device bus and traditional costs. If no price exists under a bus for a particular device, it's because a comparable device for that protocol is not yet for sale.

The materials and labor costs table captures the costs of materials such as cabling, junction boxes, power supplies, host interface, and I/O cards, as well as the associated labor charges to install and commission the system.

In the engineering time table, place all the engineering tasks associated with a project and estimated time required on a per-device basis.

Component I/O list contains a summary of the traditional I/O required by each device type. This data helps calculate traditional wiring requirements.

The engineer running the model can edit all of the values in the constants sheet. Some tables, such as the design criteria and the device cost/availability tables, are project specific and use information from local vendors regarding design and price estimates.

Others, such as the engineering time table, are much more difficult to estimate, so these values derive from other existing network economic calculators models.

BUS DETAILS SPREADSHEETS

The bus details spreadsheets are five separate sheets: one for each bus type, plus one for traditional I/O. Their purpose is to provide a main calculation area where the specific materials costs and design intricacies of each bus can be accounted for.

In particular, the hardware determination table is where bus constraints such as the number of devices on a network or spur; the number of devices per terminal assembly, junction box, or power supply; and the number of networks per host interface card are set. In general, the user does not edit these values.

The other options on the worksheet enable the user to alter the labor and material unit costs for the particular bus system if desired. The remainder of the worksheet calculates and summarizes the design and commissioning costs of the individual systems.

The inputs and results summary worksheet serves as the main user interface, where actual device counts assign to each fieldbus system to determine the most economic overall installed cost.

At the bottom of the worksheet is the results table, where the comparative and total costs appear for the combination of devices entered in the upper part of the worksheet. Subtotals present the device premium, which is the incremental price of the fieldbus device vs. a traditional analog or digital device; the materials cost of conduits, trays, terminals, cabinets, and the like; the installation labor costs; the design costs, including the engineering and project management expenses; and the commissioning costs associated with unit start-up.

The intention of this main sheet is to allow the user to run a number of different what-if options to determine the cost impact of assigning different devices to different buses.

In fact, by using either Microsoft Scenarios or a Visual Basic program, the system can search for an optimal bus combination for a fixed set of devices.

Based on the device counts provided by the company control engineers for a typical food plant, we ran a number of scenarios for different bus combinations.

Device availability constraints and cost calculations quickly narrowed down the possibilities to an FF/DeviceNet combination or an FF/Profibus-DP/AS-i combination.

For comparison purposes, we analyzed a traditionally laid-out system, too.

In the final analysis, the choice of buses has relatively little impact on the overall project cost. The combined cost of materials, labor, and commissioning is very similar for each of the four buses.

The only noticeable impact may be on the engineering costs and on the maintenance costs, as additional training and spares will be required for each new bus.

However, the net result when considering the total installed cost is that a mixed fieldbus system will result in project savings of about 30% over a traditional, hard-wired control system.

AND THAT AIN'T THE HALF OF IT

Understand that the above model provides information on the key cost variables at the start of a project, not the long-term costs or savings once the project is complete.

Understanding these project costs are necessary to convince a cost-conscious project manager or approval committee. The model does not yet consider the fiscal benefits over the life of a project. We searched the literature and interviewed bus systems users.

Fully 80% of the cost associated with any industrial purchase occurs after it has installed. If a device has a life cycle of 10 years, maintenance reduction related to that device will be significant.

Interviews with users in 10 different industrial facilities indicated that all fieldbus systems reduced maintenance costs because they incorporate some level of device diagnostics. Even AS-i has this capability, though the level of detail is limited.

Device maintenance information and bidirectional communications make it possible for maintenance technicians to determine the "problem" with a device before going to the field and, in some cases, even negate the need to go into the field.

A further study indicated a savings of 20% in maintenance support and 40% in inventory from using fieldbuses. Fisher-Rosemount reported a study showing that in one large chemical plant, 63% of the trips to the field to check transmitters were unnecessary.

An additional 20% of the checks resulted in calibration changes (span and zero). With digital communications, up to 83% of these field trips would disappear with the use of the integrated, computer-based maintenance software on the host control system.

In most sites, this would result in a 15% savings on instrumentation maintenance costs.

LIVING LONG-TERM TOOL

All major instrument manufacturers and host system suppliers are migrating their products to support fieldbus and digital networking. As a result, by the end of the decade traditional analog devices will be a niche product-the most expensive systems to maintain.

Such systems and devices will likely require the use of handmade boards because the chip sets on which they sit will no longer be available.

Fieldbus systems, on the other hand, will be the norm, and because they are "all digital," it will be possible to keep them current by simply installing the latest software upgrade. This is quite simply the best way to get the greatest benefit at the lowest price.

Control systems have a life cycle of typically 10 to 15 years. Does it make sense to start a project today that will obsolesce, if not by the time it commissions then partway through its life cycle?

As Fisher-Rosemount's Duncan Schleiss opined at a recent conference, "The challenge faced by host suppliers today is that hardware [chip sets] and operating systems have a three-year life, while support for these systems is 10-15 years."

Digital networks protect this investment to some extent. As long as the CPU and memory are sufficient, the controlling software can change via download to devices.

Probably the greatest opportunity lost by not using a digital control system is integration of information throughout the enterprise. If knowledge is power, then digital control systems are the path to power because they make it possible for an organization to know the process right down to the sensor level. IT

Behind the byline

Ian Verhappen is an engineering associate at Syncrude Canada and chair of the Foundation Fieldbus End User Advisory Committee. He is a senior member of ISA. Eric Byres is research faculty and manager of the Internet Engineering Lab at the British Columbia Institute of Technology and is also an ISA member.