This article was printed in CONTROL's September 2009 edition.
By Adel Ben-Duheash and Mohammed Batouq
As the world leader in crude oil production, Saudi Aramco's operations span the globe and the energy industry. We run an extensive network of refining and distribution facilities and are responsible for the oil and gas processing and transportation installations that fuel Saudi Arabia's industrial sector, thus contributing to Saudi Arabia's status as the leader in petroleum exportation.
As our company's largest oil processing facility and the world's largest crude stabilization plant, the Abqaiq Plants facility plays a pivotal role for Saudi Aramco with our crude oil production capacity of more than 7 million barrels per day. Such extensive and sophisticated oil exploration, production and exportation requires Saudi Aramco to use cutting-edge technology because even a very short period of unforeseen downtime can costs millions of dollars.
Fortunately, with help from system integrator Integration Objects' OPC products and integration services, our process, data and alarm management technologies at Saudi Aramco in general and at Abqaiq Plants in particular are second to none.
Time for a Change
Complexities and diverse needs are expected for any large oil producer with significant plant operations. Such complexities are enhanced as plant operations grow, as technologies get updated, and as the needs of the plant and business systems evolve.
With this in mind, Abqaiq Plants' engineering personnel monitored the industry's technology trends, and sought an open, proven technology that could protect its infrastructure and pave the way for future expansions. Many alternatives and products were tested and piloted to find the most feasible and practical approach to align Abqaiq Plants' infrastructure with the rapid advances in the industry. Saudi Aramco Abqaiq Plants had a set of stringent and complicated requirements to assure a continued and uninterrupted flow of process data. Integration Objects was selected because met the rigorous and complex requirements that Abqaiq Plants had to integrate a 23-point-to-point connection infrastructure based on the VMS and Unix platforms that connected all the distributed control systems (DCS) and other automation islands with the plant information management system (PIMS).
It is not unusual to find such complex architectures in large plant systems where operations and, therefore, necessary support structures expand over time. For Abqaiq Plants, even though each point-to-point connection addressed a particular need, the overall system architecture was cumbersome and costly to update and maintain because, by then, VMS had been practically fazed out of the industry, so interoperability and spare parts for the hardware became hard to find. Changes in the DCS, such as upgrades and replacements, affected the communication between applications, and each change also typically required implementation of new point-to-point interfaces. This made us at Abqaiq Plants realize that each point-to-point interface required specific maintenance, spare parts and maintaining on-site expertise, which added considerable costs to our operating budgets. Despite the complexity, we were determined to explore and pilot all technologies and useful applications that could help optimize our day-to-day operations.
Once we decided to upgrade and improve the efficiency of our DCS interconnections, and after a thorough evaluation and testing of many products, we found the solution from Integration Objects. Our primary requirements were:
- Use industrial standard solutions
- Simplify the overall system architecture
- Implement a more flexible architecture that could be easily adapted for future changes in the plant control systems
- Improve reliability and robustness
- Migrate plant MIS from Unix and OpenVMS to Windows.
An Integrated Solution
Using these guidelines and considering the needs and complexities of our specific operations, Abqaiq Plants' engineering team, system integrators and project engineers from Integration Objects implemented a solution with the final following architecture:
OPC was identified as the industrial standard solution able to simplify the whole system architecture. So, each DCS exposes its variables on an OPC server provided by the DCS manufacturer, and data exchange is performed using OPC clients that access data on the servers. The OPC client-server approach also creates a more flexible system that is useful in case there are any future changes.
To simplify the system architecture, the engineers installed a redundant hardware configuration by running a virtual infrastructure (VMWare virtualization software). Two virtual servers were deployed on a redundant configuration with each hosting 10 different virtual machines—one for each OPC server. These servers were configured as two units, where the first acts as a primary machine and the second (geographically located at another site) is a backup.
Because OPC servers can't directly communicate with each other, inter-DCS data exchange was provided by installing Integration Objects' OPC Data Transfer, a plug-and-play software tool for transferring real-time data. OPC Data Transfer fixes the problem by configuring and monitoring bi-directional data transfers between OPC servers, so all processes are integrated.
Communication between the plant historian, OSIsoft PI, and the DCS' OPC servers was based on OSIsoft's PI-OPC interfaces. These consist of OPC clients that are able to manage the redundancy of the OPC servers and their switchover.
Redirection of clients to the appropriate server is done automatically. At this point, it was necessary to ensure the clients' redundancy to guarantee that they can't become single points of failure during daily operations. This need was satisfied by installing them on servers running Marathon EverRun software. This software synchronizes two Windows servers to create one application environment that runs on both servers simultaneously, and maintains availability under any circumstances.