Predicting risks: how predictive analytics will change the oil and gas industry

Predicting risks: how predictive analytics will change the oil and gas industry

It becomes more and more difficult, from year to year, to ignore the technological paradigm. Digital transformation encompasses, to one degree or another, basic infrastructures and thus extends to many industries. Businesses actively introduce a digital toolkit allowing them to increase production efficiency, estimate possible risks, precisely plan repair campaigns, and optimize logistics.

Predictive analytics and remote monitoring systems are one such tool. In energetics, PRANA — a system manufactured and successfully adapted to meet the needs of the oil and gas industry by ROTEC — has become the industry standard.

ROTEC
15 Nikoloyamskaya St., Moscow, 109240, Russia
+7 (495) 644-34-60
28 October 2021

It becomes more and more difficult, from year to year, to ignore the technological paradigm. Digital transformation encompasses, to one degree or another, basic infrastructures and thus extends to many industries. Businesses actively introduce a digital toolkit allowing them to increase production efficiency, estimate possible risks, precisely plan repair campaigns, and optimize logistics.

Predictive analytics and remote monitoring systems are one such tool. In energetics, PRANA — a system manufactured and successfully adapted to meet the needs of the oil and gas industry by ROTEC — has become the industry standard.

Transition challenges

When introducing the system into the oil and gas industry, the developers faced a significant problem: varying process and operation conditions of equipment between different industries. Power machines run in steady-state mode most of the time. Operation of turbines and other machines used in the oil and gas industry includes many more transition modes with fast starts and stops. This means that monitoring systems that mainly use statistical mathematical models to monitor the equipment condition will not be quite efficient for some oil and gas facilities.

0ac18879ca4eb55ca59ce94f5

Example of how a statistical model works in an abnormal situation. Increasing vibrations, though technically remaining within the permissible range, still lead to a rise in the integral condition criterion of the machine. Combined with other model parameters, its behavior is unusual and indicates an abnormal technical condition.

As a solution to the problem, the developers have suggested that a hybrid digital model, where statistical mathematical modeling is combined with physical mathematical models, should be used for equipment operating in extreme modes. According to the developers, this will allow:

  • more precise prediction of possible emergency situations;
  • identification of a specific defect cause with a high degree of probability;
  • more efficient recommendations as to further system actions.

To test the hypothesis about the practical effect from the combination of two different methods, ROTEC has developed a new PRANA-based prototype at the premises of a gas transporter.

Hybrid model: how it works

As expected, the hybrid model made it possible to encompass most operating modes of equipment, including transitions. With their different approaches to evaluating the machine condition, statistical and physical models have supplemented each other, dramatically improving prediction accuracy.  

In steady-state modes, excellent convergence has been demonstrated by the statistical model that is a kind of the reference digital portrait of equipment. It relies on a set of data about normal operation of a machine in various modes. The system compares current and ideal parameters and displays the varying technical condition of equipment in real time.

This method is best suited to long-term forecasts as it allows the overall condition of equipment to be evaluated and the trouble to be localized where possible. 

ab6270e0dcafd5510426f878b

Interface example showing the physical model parameters of a gas generator and power turbine within a gas compressor unit.

The physical model has provided a more detailed picture of machine operation over short time intervals (within ten minutes), primarily under extreme loads. The principle of the method is to simulate the operation of equipment used in a process. The model is cyclically reproduced over time, allowing reference parameters of equipment components to be calculated for various modes. The obtained calculations are readings from virtual sensors that are generated by the model. 

Divergence between actual parameters and those estimated by the physical model allows a clear determination of which component has malfunctioned and identification of a specific defect cause. As an additional benefit, the method also offers the possibility of simulating machine behavior with an induced fault in order to analyze further behavior of the system with a similar defect. In the long run, this will prevent accidents related to habitual equipment failures.

The prototype of the hybrid monitoring and predictive analytics system already runs on six gas compressors installed at three gas distribution stations. This is the first practical step toward the comprehensive transition of the oil and gas industry to advanced predictive analytics and monitoring systems. A new tool that will help prevent emergency situations and greatly increase business efficiency through decreased equipment downtime and reduced repair expenses.

Source: Energy and industry of Russia
See also
Make your first step in the digital economy
Leave your phone number and our expert will contact you for consultation
Your phone number *

© PRANA: Predictive Analytics and Remote Monitoring System, 2022

15 Nikoloyamskaya St., Moscow,
109240, Russia