Invensys User Group

Use cases of DynSim and Olga include topside modeling, from design through training and operations. PipePhase and NetOpt used to optimize ESP performance. Stan de Vries explains why ‘most integrated asset management projects fail.’ Production optimization on Petronas’ Baronia.

Cal Depew (Invensys) described how DynSim and Olga are used across the facility lifecycle from integrated subsea pipeline and topside modeling, through DCS checkout against dynamic models and on to training and operations. This is a rich field for front end design and optimization—investigating issues such as how topside operations impact pipeline design in terms of slug formation, produced water handling and flow assurance.

Alexander Chamorro showed how electrical submersible pump (ESP) operations are optimized using PipePhase and NetOpt. Chamorro noted that 60% of the world’s oil wells require artificial lift and of these, 14% use ESPs. The ESP is perhaps ‘the most versatile and profitable piece of equipment in a petroleum company’s arsenal’ but ESPs can become an expensive nightmare if not properly operated. Hence the need to optimize ESP performance to maximize run life. ESP sizing is key at design time and many factors such as drawdown and motor load must remain in an optimal range—in the face of changing conditions over the pump’s lifetime. Chamorro noted that ‘most complaints regarding pump performance stem from placing a pump in an application that requires it to operate outside its optimal flow or pressure ratings—causing for example gas lock or pump-off.’ ESP optimization is achieved by detecting deviations from established trends—and acting on them a.s.a.p. to reduce risk of pump failure. Invensys’ NetOpt was demoed on a gathering system, using the objective function to maximize flow rate through the network, while keeping pump rate, head and motor horsepower in range.

CMG VP Jim Erdle showed how the STARS reservoir simulator, coupled with the PipePhase surface network simulator is used to optimize steam assisted gravity drainage (SAGD). This high cost, low margin exercise mandates optimization. The objective function is designed to maximize NPV in the face of the ‘competing’ objectives of producing more oil and reducing steam consumption.

Invensys’ Stan de Vries claimed that most attempts at integrated asset models (IAM) fail because there is not automated application and data management. Global optimization is different from local optimization and requires a move up the data/information/knowledge stack. Unstructured control system data needs processing for event recognition and transforming into actionable info. Along the way, we need to automate data quality management, adding-in virtual metering and data reconciliation. Fortunately, moving up from local to global optimization, taking account of well interactions, results in a 4-5x speedup in convergence. de Vries illustrated this with a case study of hydrate formation in a gathering system. Here the software computes phase conditions, triggered by temperature difference and flow reduction—and suggests a fix. The spin-off is that this can be used to put a monetary value on the digital oilfield approach.

Brian Dickson offered a more prosaic digital oilfield example using the Invensys/Foxboro digital Coriolis dual phase measurements for well test, CO2 and water injection management. This hardware produces a massive data set which was plugged in to a neural net solution for wet gas monitoring, allocation and reservoir optimization.

Harpreet Gulati showed how production optimization is used to address problems such facilities bottlenecks. Such techniques can be used even in the face of incomplete information from the field. Gulati showed how a combination of PipePhase and Romeo in an automated asset model can provide key performance indicators and even generate new setpoints for enhanced operations. PipePhase is used to model complex and extensive networks, including constraints. Network models can be combined. A study by Genesis Consulting validated the Romeo/PipePhase approach. Invensys’ Romeo online optimizer is used by ‘most major oils’.

Gulati then turned to a case history of a process optimization advisory developed for Petronas’ Baronia Platforms. This was used to optimize production and minimize gas venting, leveraging installed IT and the Foxboro/Invensys DCS. Again a ‘rigorous’ PipePhase/Romeo IAM was developed for Baronia’s 21 wells and 4 production manifolds. The Romeo facilities model downloads data in real rime, performs data reconciliation and model tuning. The model then back calculates individual well production which is used for optimization. The IAM considers all process interactions to optimize choke settings, gas lift rates, separator pressures and compressor suction. All of which is done while considering constraints such as gas venting, gas water dew point, gas and oil export pressure and so on. Sub components of the extensive Baronia IAM include a real time system model, a steady state detection model and a model sequence activation controller.

The Baronia IAM performs extensive data management (as proposed by Stan de Vries above), leveraging Invensys’ InSQL historian to capture operating data, well status, price data etc. Operator and engineers review and then implement the advisory set points. Petronas uses the IAM as a motor for a continuous improvement loop around data reconciliation, optimization, execution (set points) and so on.

Gulati concluded saying that information—especially oil and gas compositional data are pre-requisites in developing a reliable model. The IAM resulted in a $40,000 per day revenue hike, reduced gas vent, improved production allocation and compressor performance monitoring.

Hesh Kagan’s presentation (with Motorola’s June Ruby) described a ‘wireless win’ in the refit of a refinery. Wireless was deployed to connect remote tank farms. PLC’s in each farm converted gauge data to Modbus over Ethernet—with dual wireless paths for redundancy. At the pump house data was converted back for consumption by the legacy I/O modules.

Larry Balcom presented more work performed with CMG, on SimSci/Stars integration—again in a heavy oil context. In a horizontal well, PipePhase models what happens inside the liner while CMG’s Stars models what happens from the liner to the reservoir. A new interface is under construction to map from PipePhase to Stars, integrating with Sim4Me. The idea is to be able to run SAGD simulations across heating, water flood, steam injection, bottom hole reactions and refining capacity. A simulation executive allows models to be scripted and produces AVI movies of simulator results. More from

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.