When to Upgrade, Migrate, or Replace

Improvement of control systems (such as PLCs, DCS, HMIs, and peripheral systems: drives, analyzers, and gauging systems) involves two distinct types of activities. One set of activities can be performed off-line without disruption to production, while the other is done in-situ and requires shutting down the production equipment.

By David St-Onge, Boyle Outsourcing Inc. August 1, 2004

AT A GLANCE

In-situ and offline options

I/O field signal conversion

Sectional switchover

Making the choice

Sidebars: Control systems replacement activities

Improvement of control systems (such as PLCs, DCS, HMIs, and peripheral systems: drives, analyzers, and gauging systems) involves two distinct types of activities. One set of activities can be performed off-line without disruption to production, while the other is done in-situ and requires shutting down the production equipment. The main tasks associated with off-line and in-situ activities are listed in the ‘Control systems replacement activities’ graphic at right.

The in-situ field activities result in significant downtime and production loss. A typical industrial production facility might replace an old control system for a production line composed of the 50 units of operation, 60 motors, 900 field devices, and 2,000 field I/Os. Depending on whether the motor control is performed in a PLC or directly in the DCS will affect the downtime calculation by about 15%. Four technicians per shift perform work over two eight-hour shifts. Usually I/O rack rooms are small and trying to add more people will be counterproductive. Typically a pair of technicians can check approximately 15 signals per eight-hour shift. Regular hourly rate is: $50/hour and overtime rate is $75/hour. We are assuming that there is Marshalling cabinetry in the I/O rack room and that new cables have been run between the cabinet and the new I/O racks and that cables were tagged before the shutdown. (See ‘Production losses …’ graphic for I/O replacement details).

One could argue that staggering the above tasks could bring down the total downtime to 35 days, but this will not allow for unforeseen problems common in replacement of old systems. In any case, 30-plus days of downtime for a budgeted production year of 360 days results in a net loss of about 10% in production revenue plus profit in addition to approximately $200,000 in I/O replacement and checkout cost.

That is a high price to pay; and even if there were several valid reasons favoring replacement of a control system, benefits would rarely justify costs. The dilemma between enduring production losses and the desire to replace a control system often persists for several years until a major refurbishing-modernization or expansion project is announced. Major work entails production downtime allowing for control systems upgrade or replacement.

I/O conversion

Astute system developers have come up with the idea of tying a new control system to the existing I/O without touching a single wire. It has opened the doors to less costly system upgrades, migrations, or replacement with minimum disruption to production.

The ‘I/O field signal conversion’ graphic illustrates the concept of I/O field signal digital conversion and storage to memory registers. The incoming field signals are converted at precise time intervals to digital values stored in memory registers. These values are, in turn, processed at precise time intervals by programs and algorithms, resulting in values stored in memory register which are sent back to field to start, control, monitor, or stop equipment and devices. The two main methods of tying into an I/O card consist of either using a physical connection or a communication interface.

Handling of control-loop I/O field signal digital conversion and memory register storage are among migration/upgrade considerations.

Physical connection —Two types of physical I/O can be connected to a new system without altering the terminations: a) card cage connection —fabricating a card cage or a card-edge connector that will accept a competitor’s I/O card. Before implementing, make sure to try each card type (AI, AO, DI, DO, iDI, iDO) on the existing, running system. Check the power, polarity, color coding, neutrals and grounds; b) swing arm connection —field wiring is attached to a swing arm that connects to the electronic card. Replacement of the card consists of lifting the swing arm and inserting a new card and reconnecting the swing arm.

Communication interface —This type of connection consists of tapping into a competitor’s I/O card communication signal and converting information to the format of the host system.

In both cases, special care must be given to ensure that the following are maintained:

Initial I/O diagnostics;

I/O programming functions;

Power supplies; and

Grounding integrity (optic isolation).

Avoiding the need to disconnect and reconnect wires creates savings. Furthermore, today’s systems come with simulation packages that allow checking the software configuration from I/O memory register.

Often, the checkout of the ‘I/O piggyback physical connection’ will cost 25% to 50% more, as it can require that I/O addresses be renamed to an intermediate register address to perform simulation and renamed to its initial address once the simulated checkout is complete.

One-shot vs. sectional

In the previous example, the switchover was performed in a three-day shutdown. Such a switchover of the entire system in one shutdown is challenging and may not always be practical. The confidence level depends on good project management capabilities, staff availability for planning and execution, staff experience, engineering support, rigorous planning, process complexity, safety, and production pressures. If it is unrealistic to perform the switchover in one shutdown, then it must be done in sections—making use of regular, shorter shutdowns.

The production line must then be divided into self-contained sections, which are usually delimited storage tanks or a holding area. Generally, less control information passes between self-contained sections. This minimizes the amount of critical interlocks and signals that need to be hardwired between old and new systems during the transitions. Also, when the inevitable startup challenges begin, holding areas allow more time to fix problems.

This approach takes longer in calendar time and in absolute working hours to plan and execute. Existing documentation must be up to date. Hardwiring of signals is required before each sectional switchover, often resulting in short circuits or grounding problems, which can shut down production. Coexistence of two systems extends several weeks and is confusing for operators, who will need around-the-clock support during that period.

Choices

Whatever the reason for improvement, the choice depends on two factors: How much production downtime does the budget allow for the switch over? What is the budget? The answers to these questions will help determine which of the following paths you choose:

Buy new or replace. It is always best to install a new control system. The P&IDs are well developed, the operating and control strategies are understood, the block logic diagrams and loop sheets are clear. Integration with other plant systems is easier and offers more value. Control system replacement as part of a larger project can offer ample time to accomplish the work. It is also highly recommended to replace and install new when the old system has been patched with add-ons (and quick fixes beyond repair) without adequate documentation.

When budgets are low and time is not available, however, upgrades or migrations are excellent options.

Upgrading means keeping the same I/O, usually staying with the same vendor, and converting the system to the most current version of the same architecture. Vendors even subsidize this type of conversion. It’s a good business decision as the client stays with the vendor and it allows vendors to phase out older systems.

An upgrade will usually come with utilities that translate the existing configuration and convert the existing graphic pages. This is less costly. Beware, however, because some converters only emulate old graphics and cannot be opened by the new system’s graphic editor. Also, since these changes usually come with increased I/O count, there is plenty of work remaining.

Though desired and useful, upgrades can result in hardship. For a few weeks, some unforeseen system behavior can occur, such as functions ceasing to work, erroneous automatic reports, and production stoppage may occur.

Migration means piggybacking to the existing I/O connections without altering the terminations and migrating to a new system architecture (same or new vendor).

The motivation to migrate typically stems from the following factors: Dissatisfaction with present vendor, concern with vendor viability, need to implement advanced features, implementation of information management systems (for example, asset management, MES, LIMS), new fieldbus, advanced controls, documentation (loop configurators that generate loop drawings), valve signature, data reconciliation, trending and archiving, workflow and documentation of configuration modifications, and inventory control.

Migration to a new architecture is challenging as functionalities differ not only between architectures but among vendors. Even with automatic software converters, functions need to be reviewed and some reengineered. (See this article online for information about the programming involved in migration despite the presence of converters). Any migration or upgrade also offers opportunity for process streamlining and procedure modification. With few exceptions, old workflows can be made more efficient by applying today’s knowledge and methods.

The ‘Decision table’ graphic provides a guide to decide the route to take when the day comes to improve a control system. The determining factors are: the state of the existing system, time available to make the switchover, and budget.

Whatever the decision, a control system will hold the fate of the entire production for the next few decades. It must be given serious consideration. Consequently, do your homework, plan in advance, and write rigorous specifications. Review existing documentation and update it. Don’t hesitate to seek advice from experts.

Production losses and cost of I/O replacement and checkout

Tasks
Days of downtime
Labor cost

Disconnecting & reconnecting wires
5
$20,000

Checkout of signal integrity
33
$163,200

Total
38
$183,200

Comparisons between ‘traditional I/O replacement’ and ‘I/O piggyback’

Traditional I/O replacement method

I/O piggyback physical connection

I/O piggybackcommunication interface

Tasks
Days of downtime
Labor cost
Days of downtime
Labor cost
Days of downtime
Labor cost

Disconnecting/ reconnecting wires
5
$20,000
0
$0
0
$0

Checkout of signal integrity
33
$163,200
0
$75,000
0
$50,000

Commissioning and startup
3
$32,000
3
$32,000
3
$32,000

Total
41
$215,200
3
$107,000
3
$82,000

Decision table: upgrade, migrate, or replace

Available switchover time
Well documented Few add-ons & quick-fixes
Poorly documented Many add-ons and quick-fixes

Note 1 : These changes, with limited switchover time, need to be done in several segments, during regular scheduled maintenance shutdowns. They involve simultaneous operation of the old and new systems, typically take several months, and are demanding on the personnel.

State of the existing system

Weeks
Low budget: Migrate
Low budget: Upgrade

Substantial budget: Replace
Substantial budget: Replace

Days
Low budget: Upgrade
Low budget: Upgrade

Substantial budget: Migrate
Substantial budget: Replace

Programming checklist for control system upgrades, migrationEven though control system vendors offer code converters, further programming is likely to be required. Following is a rundown of the most common programming issues faced when migrating to a new system:

Function blocks will behave differently. For example PID blocks have hundreds of configurable parameters such as “BUMPLESS_TRANSFER,” “OUT_TRACK,” “SPLIT_RANGE,” and “RATIO_CONTROL.” Each major vendor designs these differently.

Programming Language. Depending on the vendor, the programming language will look like Assembler, Pascal, FORTRAN, or Basic. Similar statements like “IF_THEN_ELSE” may behave differently.

Sequential programming. Each vendor has a sequential language package. Again, the features and how they behave vary widely. For example, the “CALL,” “GO_TO,” “WAIT,” and “RESUME” statements must be understood before the existing software of the previous system is converted. The debug utility that takes a programmer step-by-step is markedly different from one system to another. This is important as it determines how one designs the software structure.

Operator graphics. As a rule of thumb, a new graphic page (10 motors and 30 dynamic field elements and pages links) using the standard system functionalities goes through the following stages of construction:

Design, drawing and review 8 hours

Editing of the static drawing 8 hours

Editing of the dynamic elements and tag assignment 8 hours

Checkout, commissioning and startup 2 hours

When dealing with a migration (unless the converter proves outstanding in tests), it is better to redo the graphics. The design, roughly 30% of the work, is already done as the existing graphics are mature and have long been optimized. Even if the converter was of high quality, differences in functionality mentioned above will impact the graphic representation, requiring significant rework.

Related reading: “8 Ways to Improve Control System Projects” “Time to Reinvest in Automation”

Author Information

David St-Onge has more than 25 years hands-on experience in production, automation, and control of industrial processes and presently heads the project management division of Boyle Outsourcing Inc., headquartered in Montreal, Canada;

Control systems replacement activities

Off-line activities

P&I reviews

Block logic reviews

Operator graphic review

Alarm management review

Control system configuration

Verification of control system functionalities

Configuration checkout using simulations

In-situ field activities

Disconnecting & reconnecting all the existing wires to new I/O cards

Checkout of signal continuity and integrity from the field to the system software

Commissioning

Startup