We carry out research and development projects co-financed
by the National Center for Research and Development

POIR.01.01.02-00-0060.15 – Implementation of innovative methods to increase security and minimize losses in the transmission of crude oil and petroleum products
Project objective:
The aim of the project is to reduce pipeline protection costs by developing new leak detection methods based on cathodic protection systems and to improve existing methods by eliminating false signals through the use of artificial intelligence algorithms.
Objectives of the prototype plant:
- Reliable verification of leak detection methods used on existing fuel pipelines.
- Verification of method effectiveness and popularization of the idea of pipeline protection by automated methods
Analyzed will be innovative solutions consisting in:
- The use of artificial intelligence to filter events from measurement systems
- Implementation of a new method of leak detection based on the analysis of measurements applicable in cathodic protection
The aim of the research work carried out when constructing the plant is to develop leak detection methods that will eliminate or minimize the disadvantages of the current solutions
- high cost of the systems, which makes it impossible to use them in smaller plants,
- low reliability of detection, related to the lack of intelligent algorithms filtering the obtained measurement data
- the need for complicated and costly measuring equipment
Tasks carried out under the project
- Construction of the plant allowing for testing methods of detecting pipeline leaks, verifying and demonstrating their effectiveness to a potential recipient of solutions
- Reliable verification of known leak detection methods
- Development of new, innovative methods
- Promoting the idea of protecting fuel pipelines as an effective tool enhancing investment and environmental security
Development of an IT model of a fuel aging monitoring tool based on a numerical analysis of physical phenomena
Aim of the research
Defining an effective aging model for liquid fuels and defining an IT tool for monitoring these processes.
The aim of the project was to create a model that would allow for forecasting the moment when the fuel becomes out of use and to describe the relationship between changes in IR spectra of an aging fuel. An additional goal was to construct a fuel blending calculator, used to calculate the values of parameters on the basis of the knowledge of the values of the parameters of the components being blended, taking into account the compliance of the output fuel with the standard.
The problem of fuel aging
During storage of liquid fuels, constant changes in their composition and physico-chemical structure occur. All of these time-dependent changes are called the aging process.
The ability to anticipate changes due to aging would significantly optimize and simplify fuel storage and management processes.
Preparation of data
A set of hypothetical parameters prepared by the he Air Force Technical Institute was used in the research. The obtained sets constitute the so-called “ideal” waveforms, i.e. devoid of measurement errors.
Expected results
- “Real data” from the obtained waveform by interpolation of intermediate data using the spline method
- “Measured data” based on “real data” and the specified measurement uncertainty of each parameter.
- “Processed data” on the basis of real data and measured data after using functions modifying the domain and values
Approximations
The simplest class of models will be models based on approximation of relevant parameters. Linear approximation model for forecasting parameter changes
- Linear approximation of parameter dependence on time
- Linear approximation based on n-recent measurements
Approximation model for non-linear functions.
- Finding function classes for each parameter
Neural Network Model (ANN).
- Building test networks. Checking the information capacity of the network, depending on the adopted model. Testing the ability of networks to learn and overfitting. Checking the quality of interpolation and extrapolation of the network, i.e. its response to input data outside the range or between the data used to teach the network.
Classification
An alternative approach to the problem of predicting fuel quality changes is to use classification methods instead of extrapolation of quality parameters. This approach does not allow predicting the exact values of individual parameters in the future, but only a general assessment of the suitability of the fuel after a certain period of time.
As part of the research, a classification model based on the Support Vector Machines (SVM) was developed.
Analysis of IR spectra
One of the measurements carried out on liquid fuels is the measurement of the oscillating absorption spectrum in the IR range.
The aim of the research was to provide tools allowing for organoleptic or automatic capture of fuel quality changes on the basis of changes in its spectrum.
This model best illustrates the physical and chemical processes taking place in fuel.
The IR spectra obtained were visually analyzed in order to capture the type of changes and track them in a computational way.
Fuel blending analysis
During fuel storage, some of its parameters will be out of standard. In order to restore aging fuel to usability, the most frequently used procedure is blending it with fuel of proper parameters. The process of blending and the resulting data were subjected to research analysis.
Result of the research – IT tool design
The aim of the research was to confirm the possibility of using modern IT methods for modeling physicochemical phenomena occurring in aging fuels. Positive verification of some of the methods allowed to define basic assumptions for a tool helping to manage fuel storage facilities.
The model of an IT tool for monitoring fuel aging process based on a numerical analysis of physicochemical phenomena will be a part of the fuel supervision system.