Valve monitoring via IIoT supports Petrobras’s decarbonization program

Brazilian national oil and gas company, Petrobras, produces 2.7 million barrels of oil equivalent daily from 56 production sites and rigs, with the product refined at 11 facilities. Like many companies, Petrobras has set environmental goals across all operating areas, including reducing absolute operational emissions by 30% compared to 2015 levels with a deadline of 2030. It also has a long-term goal to become net-zero by 2050. As is common in many companies, it’s necessary to analyze which technologies and work practices are necessary to achieve these goals. 

One major area is flare systems, which are essential safety systems in refinery operations. During imbalances between gas production and its transformation into products, flare systems relieve excess pressure and prevent equipment from over pressurizing. They convert process gases—such as hydrocarbons—into compounds with less environmental impact in the atmosphere, such as carbon dioxide.

These sensors enable the identification of PRV releases, either during system protection events, or in cases of leakage caused by malfunctions. The sensor operates at ultrasonic frequencies, and detects the turbulent flow created when fluids pass through a nozzle or other restriction. This is the same principle used in PRVs, where acoustic sensors pick up high‑frequency turbulence as the valve opens or leaks. Because steam traps also create turbulence at internal orifices, ultrasonic sensing can also be used to determine whether they’re closed or passing through.

The same sensors can also be installed on steam traps to determine which are functioning correctly compared to those likely stuck closed (not draining condensate) and others stuck open (releasing live steam). This project at Petrobras was called LeakSpy (Figure 3) because any PRV or steam trap could be monitored remotely and continuously. These types of sensors can also detect changes in pipe temperature, and support operations teams performing leak analyses.

Implementing the concept

While it sounds like a simple concept, initial implementation of Rosemount 708 wireless acoustic transmitter was more complicated. Refineries are, and must be, highly conservative when evaluating new technologies because there’s so much at stake, especially regarding personal, environmental and operational safety.

Implementing LeakSpy therefore required an eight-month process of evaluation and testing prior to initial deployment. As this marked the first major implementation of the underlying wireless and Industrial Internet of Things (IIoT) infrastructure, the full project involved the simultaneous adoption of two technologies: acoustic monitoring and wireless networking.

PRV monitoring tests were carried out at Emerson’s test facility in El Campo, Texas, and steam trap monitoring tests were performed at the Pickle Center the the University of Texas at Austin. An IT architecture and cybersecurity assessment to ensure compliance was conducted by Emerson’s team in Shakopee, Minn. Finally, a three-month field test involving eight acoustic sensors was conducted at Petrobras’ Presidente Bernardes de Cubatão (RPBC) refinery in Sao Paulo, Brazil.

After clearing these hurdles, the initial large-scale deployment at RPBC (Figure 4) started with the following activities:

• Installing 120 acoustic sensors, with 114 on selected PRVs and six on steam traps, spread across six processing units in the larger refinery;
• Installing approximately 30 WirelessHART repeaters to boost network density, and support meshing capability because the site covers a large area;
• Installing seven WirelessHART gateways sitewide to capture data from all the sensors;
• Assembing three data collection and analysis workstations with supporting Ethernet hardware and Emerson PRV and steam trap app software; and
Developing a 36-month service contract with Emerson to support implementation and operation.

Get your subscription to Control's tri-weekly newsletter.

This WirelessHART system is now installed at the Sao Paulo refinery, and has the capacity to handle several hundred more sensors, so this approach is highly scalable.

Analyzing data

This system at RBPC generates huge amounts of data, so Emerson provided Plantweb Insight applications for PRVs and steam traps to collect and analyze their data, supported by  AspenTech’s InfoPlus.21 historian platform. Extensive IT networkupgrades (Figure 5) were required to support this implementation, along with other corporate network areas in the facility. 

Promising results

Once the system was fully operational, the RPBC refinery started benefitting very quickly. The maintenance and reliability site teams understood they now had far more insight into their equipment than was imaginable during manual inspections.

For example, if any of the sensor-equipped PRVs open due to an over-pressure condition, multiple teams receive a notification that a release is in progress. Operators can see the duration and severity of the release, allowing them to determine how much product is being forced to the flare. They can also look at surrounding conditions to determine if there’s an actual upset, or if the affected PRV simply has an incorrect setpoint.

Maintenance can watch what happens after the release to see if the valve closes properly and completely, or if it continues to simmer, releasing product after the incident ends. Alerts can come directly to workstations, the control room, and even as emails to specific individuals. Historical information is also easily accessed to see if a given valve experiences frequent events or misbehaves often.

This approach also applies to steam traps, also using the same acoustic sensor technology. Issues with these can be nearly as costly as a problem PRV if a steam trap is stuck open and loses expensive steam (energy, in other words), or stuck closed, allowing condensate to back up into steam pipes, fouling heat exchangers. Steam loss also has environmental implications, as steam is produced using natural gas-fired boilers.

First major win

Early in the initial phase of LeakSpy running in a refining unit, operational conditions were identified that generally aren’t detected by conventional methods: short and occasional relief events in PRVs. These result from dynamic process variations, and can be see in early indications of an increasing trend of leakage through PRVs, signaling progressive degradation of the sealing surface. 

The first condition was associated with transient pressure oscillations of low amplitude and short duration. This was sufficient to trigger relief, but difficult to capture with manual measurements or indirect monitoring instruments, as previously explained. These oscillations originated from gas compression systems operating at pressure ranges close to the PRV setpoints. 

The second condition, although theoretically detectable through periodic and recurring spot inspections using portable devices, becomes operationally impractical in plants with thousands of valves, such as the unit in question, making early identification extremely challenging.

Applying LeakSpy enabled immediate detection of these relief and leakage events, allowing for detailed analysis of the causes behind pressure variations and facilitating adjustments in operational strategies to mitigate future occurrences. Furthermore, early identification of leakage trends made it possible to investigate the root causes of sealing degradation, which was linked to previous damage and maintenance interventions performed on PRVs. 

Detecting these conditions at an early stage was essential for accurate diagnosis, and for defining appropriate corrective or preventive actions, ensuring greater operational reliability, reducing emissions, costs associated with unplanned shutdowns, and enhancing overall plant safety.

Lessons learned and looking forward

While this program has been hugely successful, implementing it took Petrobras down new paths. Using WirelessHART presented some challenges with antenna and device placement, but it was still far easier and faster to install than wired solutions. Now that the WirelessHART infrastructure is in place, adding additional sensors only requires mounting a transmitter and adding it to the existing network.

The larger challenge was simply wading through the initial data to benchmark where Petrobras was and to establish next steps. Emerson worked with our site team to establish processes needed to identify problem valves and steam traps, and lay out plans for more efficient maintenance planning. As these aspects are addressed, Petrobras’ staff expects to reduce the volumes sent to flares, saving product and reducing emissions. Staff are also using data analysis tool to discover other things, such as operational maneuvers that can be improved by avoiding PRV openings. The result is improved overall operation, increased safety and reduced emissions.