EE - Forum Styles
fo

Up-front attention eases integration tasks for Shell

Mike Bacidore

Construction is underway for the Shell Pennsylvania Chemicals Project near Pittsburgh. The polyethylene plant is expected to be operational in 2021, but only because of an organized approach for the integration of third-party communications with the plant’s DeltaV control system architecture.

There are four different engineering contractors. To be successful, we had to ensure that we provided some consistent standards on how to communicate,” said Bob Sherven, principal control and automation engineer, projects and technology, at Shell. “We have an ethane cracker engineered by one engineering contractor and two gas-phase polyethylene plants by another. The third polyethylene plant has a different contractor, and the fourth contractor basically did everything else.”

Roughly 25,000 packages—digitally communicating devices or systems—were readily networked to the DeltaV system. However, 6,000 additional packages needed to be connected using other protocols.

First, we created a spreadsheet to track the many different packages,” explained Sharath Anthanahalli, senior control systems engineer at Bechtel. “We wanted to centrally manage and coordinate.” The spreadsheet included IP addresses and Modbus addresses, as well as information on how far a package was from the control room. “We captured everything and tracked the changes during the engineering phase,” said Anthanahalli.

What’s the protocol, Kenneth?

Protocol options for soft interfaces were identified early in the project:

Modbus TCP/IP for PLCs and analyzers;

OPC DA to handle the large volume of data used for monitoring; and,

Profibus for low-voltage motor controls.

We did not want a bunch of distributed PLCs throughout the site,” explained Sherven. “We wanted a network for configuration management from a central location for all packages. We networked for software patch upgrades and antivirus updates.”

The typical Modbus network included DeltaV Ethernet I/O cards (EIOCs) in redundant configuration for Modbus TCP/IP communications together with Cisco switches and Fortigate firewalls/routers. “We eliminated 99% of the cybersecurity issues with a managed switch,” explained Anthanahalli.

We had a different design for OPC,” said Sherven. “There’s a Kepware intermediary server for translation from OPC UA to OPC DA for DeltaV.”

It wasn’t always easy to get the information for each third-party package from the vendor. “If the engineering contractor was bidding out a compressor package with a PLC, after the purchase-order (PO) award, we had to get the information,” said Sherven. “It was not a requirement for them to supply the information. The PO might go to the compressor manufacturer, but they might have subbed out the controls.”

The information was then entered into SmartPlant Instrumentation (SPI), a database tool.

Testing occurred all over the globe, and the testing-schedule spreadsheet showed where and when the testing for each package was going to happen. “Most of the time, a Shell person or an Emerson person would be there,” explained Sherven. “We had two DeltaV suitcases, one in Europe and one in North America, for factory acceptance testing.”

“If the engineering contractor was bidding out a compressor package with a PLC, we had to get that information after the purchase-order (PO) award.” Shell’s Bob Sherven explained how they tracked and managed information on 6,000 digitally communicating devices and systems that needed to be connected to a new plant’s DeltaV control system using a range of protocols.

Challenge accepted

Particular challenges came from vibration monitors, an analyzer management and data acquisition system (AMADAS), substations, steam-turbine generators, optical analyzers, a tunable diode laser (TDL) analyzer and a pressure swing absorber (PSA).

We went three separate times to Munich for the PSA,” said Sherven. “There was a high latency for getting into the DeltaV.”

The system was seeing an 8-second latency from the Siemens PLC, causing considerable back-and-forth finger-pointing. “As it turned out, firmware was the issue,” explained Anthanahalli.

Two opposing package strategies exist, according to Sherven. The first is to dedicate substantial engineering time up front during the design phase, while the second focuses little on packages during the design phase and defers to the site, so the packages can be dealt with during commissioning.

We consciously chose the first strategy,” said Sherven. “The second is one you’re more forced into.”