Jul-2015

Splitting distributed control for a refinery expansion

A refiner’s expansion plan required a second distributed control system, achieved by splitting the original system without a shutdown

PETER HELMERSSON and SVEN WINBERG
Preem AB

Article Summary

Preem, with headquarters in Stockholm, Sweden, is the country’s largest supplier of petroleum products, and is responsible for 80% of national refinery capacity and 30% of Nordic capacity. Total refinery capacity is more than 18 million tonnes of crude oil at the company’s two refineries in Gothenburg and Lysekil.

The Lysekil facility, located on the south western coast, can process about 11 million t/y, shipping the majority of its output by sea.

In the past, the refinery (see Figure 1) had been controlled by a single DeltaV distributed control system (DCS) with nearly 15 000 device signal tags (DST), 56 redundant controllers, one wireless gateway, 28 operator stations and nine servers. The company has several projects and expansions in mind, yet over time it became apparent that such expansions would be difficult with a single system. The best way forward would be to split the DeltaV system into two. The challenge was to do this without a shutdown and without disrupting production.

Consulting with Emerson Process Management, Preem decided to split the system into two systems joined by DeltaV Zones, which is an architecture designed to operate and manage large systems. Because the communication across zones was process critical, Preem chose to implement redundant inter-zone servers. Figure 2 shows two zones, each representing a DeltaV system, defined by a ProfessionalPlus (ProPlus) Station, with necessary data exchange between domains handled seamlessly by the redundant DeltaV Inter-Zone Network using Inter-Zone servers. Any operator workstation within the unified architecture can view and control the areas assigned to that workstation, regardless of domain.

The immediate questions concerned where precisely to make the split and which controllers, with their I/O and associated equipment, would be assigned to which of the two new systems.

The first consideration in this process was to avoid splitting apart any controllers that are sharing data for control purposes; the ideal would be to find a logical point where the interfaces between controllers are at an absolute minimum. Logically, the split point should fall somewhere in the middle of the facility, based on unit operations.

To aid in choosing the split point, Preem engineers undertook a study that included reference visits to other companies that use Zones architecture. The refinery is already divided into two processing areas, called Area 2 and Area 3, and it was decided to make the split between them, with each to get its own DeltaV system. It was further decided that the existing system would be used by Area 3 because it contained equipment that could not be stopped during the turnaround, and the new system would be assigned to Area 2.

Preparations
Preem commissioned Emerson Process Management Sweden to carry out the project, including planning, factory testing, implementation and site acceptance testing. In order to keep everyone up to date, meetings between Emerson and Preem personnel were scheduled at two-week intervals.

The Emerson engineers analysed the database on the existing DeltaV system, looking at the configuration to see which controllers were exchanging information with others to try to identify clean break lines − places where the systems could be split between the two systems without having to do a great deal of software modification. They were able to identify a number of these; in the small number of instances where this was not possible because controllers contained modules from different areas, communications between one destination system and the other were run through the Inter-Zone Network.

Setting up the test system
Such a profound change in the refinery’s control system would, of course, require complete off-line testing to ensure that, when implemented, the split would not produce any unexpected results that would disrupt production or, worse, create any safety hazards.

While the configuration changes to the controllers themselves were straightforward, the testing was considerably more about making sure that everything that was supposed to happen did indeed happen, and that the operational speeds were satisfactory.

Because this was a live split on a plant running at reduced capacity, one of the biggest and most important tasks was to ensure that data that was moving between controllers in the original system would be able to go through the zone server as they were split into new networks.

Emerson Sweden set up a test system at its office in Karlstad. The user configuration database from the existing system’s ProPlus workstation was exported and the relevant portions were imported to the test system ProPlus workstation (ProPlus 2), which was to become the workstation for the second DeltaV system after the split. After that, the controllers and operator stations were moved from one system to the other as required, and the inter-zone server was set up.

The next step was to run tests with a set of field instruments maintained at the test site, check out operation of the Foundation Fieldbus segments, and check the status of communications.

It was also necessary to verify whether a problem during operation would trigger a redundant inter-zone server switch-over to its standby partner − which is supposed to be completely transparent − without creating any issues for the process. A considerable effort was put into the necessary configuration changes to make this work.

Next, the factory acceptance test (FAT) took place. This was essentially a repeat of the items previously mentioned, with Preem personnel observing and signing off.

Software freeze
One week before starting the split, a software freeze was implemented. No further configuration changes could be carried out because that could cause unpredictable system behaviour when the split was implemented.

A full back-up of the system was sent to the Emerson office in Karlstad, where it was cleaned up to remove any items that were remaining in the new Area 3 zone, based on the original system. This left a clean database to utilise for the new Area 2 zone.

After several iterations and modifications to plans based on information gathered during the testing, it was time to move on to implementation.