Time to replace?

Dealing with legacy systems that are becoming obsolescent often leaves process plant managers and engineers facing some tough choices around whether to migrate gradually or carry out a full-scale rip and replace programme.

Rip and replace generally means taking the whole system out, and rebuilding from the ground up, possibly with a new automation vendor. Less drastically, migration offers a continuous evolution to replace or add parts and functionality to keep a system running.

Underlining the significance of the issue, David Humphrey of ARC Research estimates that there are currently $65 billion of automation assets installed and running in the field that are ripe for replacement.

How companies tackle this issue often depends on whether or not the process operation can commercially afford the luxury of lengthy shutdowns, said Humphrey ­ chairing a panel discussion on this topic at Honeywell Process Solutions’ EMEA users group meeting in Istanbul.

He was accompanied on the panel by Jim Anderson of Saudi Aramco, Mahdi Akbar of Equate Petrochemical Co., Keith Landells of BP, Honeywell’s Andy Coward and Paul Stewart of Marathon International Oil Ltd.

The discussion highlighted a growing need for options that enable operators to migrate up more cost-effectively than at present. This included technologies that enable systems installed in the late 70s or early 80s to work alongside current technologies.

Migration is a particular problem in the oil & gas sector, which relies heavily on legacy assets ­ and where recent advances in production technology and techniques mean outages are being constantly extended.

“Whereas, before, we were typically looking at annual shutdowns, now we are looking at [one] every two to three years,” said Stewart of Marathon Oil. “These shutdowns are also getting very congested.

We would have to look at each area of the plant to see if we could shutdown certain areas to install new systems and cut things over. It is very difficult to get the window that¹s needed to replace some of these assets.”

Stewart believes, though, that the rip-and-replace option should only be considered as a very last resort ­ not least because of the amount already invested in installing and running systems.

³For me, if you stay on top of continuous evolution and keep your systems up-to-date, you do not get into a situation where you are looking at a rip-and-replace option,² he said.

Another issue is that managers and engineers have different perspectives when it comes to considering options around migration, believes Anderson of Saudi Aramco: ³As engineers we want the latest and greatest, while management comes at it more from a cost perspective.

Greater functionality is very seldom a driver for migration, continued Anderson, who pointed out that an engineer¹s project proposals have to stand up from a Capex as well as a technical point of view.

³The two main reasons for migrating are obsolescence and reliability,² he stressed. ³New functionality and enhancements do not justify a migration or a rip and replace [project]. It¹s all about safety and business risk so that at the end of the day process engineers must hand back the decision to management.²

The background to such dynamics, though, is that many offshore assets are around 20-30 years old, so that it is often a struggle for engineers to locate replacement hardware and components for a DCS system.

³The question comes when we are at the end of support or where parts are not manufactured any more,² said Anderson. ³At that point we are looking at the vendor to see if they can do a special contract to continue to supply parts and continue to offer the technical support and training.

³If the answer to that is Œyes¹ then we will pursue that contract. That gives us another five to 10 years, and then we can look at what it takes to migrate to a new system.²

Another complexity is that migration times also differ across the various parts of the system, such as the control, the production station and the server.

As Humphrey of ARC put it: ³ You have the HMI and engineering station, the control hardware system and then the I/O. It¹s like the roots of a tree:

the bigger the tree gets the deeper the roots and the harder it is to get them out of the ground.²

The expected lifetimes for level 1 controllers and the operator stations at level 2 differ, with the latter migrated more frequently, added Akbar of Equate Petrochemical. Site-specific advanced control applications or logic, he noted, tend to cause the most technical difficulties in a migration.

³It¹s very hard to simulate some applications: it is only when you go live  that you really find out. That¹s  when you¹re really under pressure to get the plant back up and running,² commented Stewart ­ making another argument for a phase-by-phase approach to migration.

The turnover for HMIs is typically every four-to-five years, though there also many older HMIs are still employed. This can create problems when installing new software if the hardware is too dated. This, said Anderson, has led to the introduction of lifecycle management requirements in industry contracts to cover aspects such as the software updates needed to support hardware over the lifecycle of a system.

From a supplier¹s viewpoint, Coward of HPS said vendors must ³move from selling boxes and support to selling Œoutcomes¹ that cover responsibility for migrating, changing, delivering, patching and securing. So instead of buying a DCS, you are buying availability of an automation system connected to your process. That¹s a very different perspective.²

Another significant challenge to a successful system transition concerns getting the operators to learn how to navigate the adoption of a new system in a short period of time, believes Keith Landells of BP.

³It¹s not technology, not usability; it¹s the soft factor,² said Landells.

³The operators just want what they are used to.² ³If you¹ve got just three weeks to get a system changed, say to a Honeywell system from a competing system how exactly do you ensure the competence? How do you train them on the schematics and how the system looks and feels?² asked Coward.

When it comes to getting experienced operators to understand a new control system sending them to training courses is not enough, according to the BP manager. ³You do the best you can with training but still there is always going to be a push-back when you introduce new systems.² One approach, he suggested, is to promote interchange between experienced and newer operators as both have something to offer to the migration project.

For his part, Akbar of Equate Petrochemical advised bringing operational staff into development meetings to explain what is going to happen.

³Getting the [operators] involved very early in the process is really valuable,² agreed Stewart. ³If they understand why things have to change then there is more chance of them buying into it. If you just present them with a new system or a change and they have not been involved in the process then they will typically reject it.²

Meanwhile, IT departments and process control department¹s often end up running two separate parallel Œchange¹ processes during the migration process, according to the panel.

³This is very inefficient, because the IT guys don¹t know much about process control model and the process control guys don¹t know much about the IT model,² ³concluded Stewart. ³We need both to work together or teach people new skills so that we get engineers who understand both aspects.²