Published June - July 2025, Pg. 54-62
Section: Еcology and industrial safety
UOT: 504.064.4
DOI: 10.37474/0365-8554/2025-06-07-54-62
Optimization of the process for cleaning oil-contaminated soils using the extreme gradient boosting algorithm
G.S. Gasanov Dr. in Tech. Sc. - Azerbaijan State Oil and Industry UniversitySoils saturated with oil undergo various transformations depending on the quantity and composition of the pollutant. These transformations affect both the structure of the soil and the compositional characteristics of the oil. Oil production and refining enterprises are sources of environmental pollution, not only due to planned and regulated activities documented in technical guidelines but also as a result of unplanned events, such as emergencies at production facilities–especially around onshore oil wells–storage sites, oil processing facilities, and during the transportation of oil through main pipeline networks or by land transport.
Based on the amount of oil that infiltrates the soil, the degree of pollution can be categorized as weak (less than 3 %), moderate (more than 3 %), or severe (up to 30–40 %). When contaminated with oil, the normal functioning of the soil system is disrupted: oxidation-reduction processes change significantly in intensity and direction, agrophysical and agrochemical properties deteriorate, and microbiological activity indicators decrease. This leads to a loss of fertility, rendering the land unsuitable for agricultural use and contributing further to environmental pollution.
The displacement of oil from the surface of soil particles by a water-washing solution is a complex process influenced by factors such as the granulometric composition of the soil’s mineral solid phase, the structure of the pore space, the degree of dispersion of aggregates and microaggregates, as well as the surface properties and chemical composition of both the oil and the washing solution. A number of additional factors also play a role.
Given the above, the study of effective technologies for cleaning, restoring, and reusing oil-contaminated soils is both scientifically and ecologically important. This work presents the results of a study on weakly contaminated soils, specifically examining the relationship between the efficiency of cleaning oil-contaminated soils in the Balakhany region of the Absheron Peninsula and various factors. These include the composition of components in the water-washing solution and the duration of mechanical action. A machine learning model was developed to analyze these relationships.
The degree of soil cleaning was used as a criterion to evaluate the influence of technological parameters. Experimental data were processed using statistical methods, forming the basis of a mathematical model built on the extreme gradient boosting algorithm. This machine learning method is characterized by high computational efficiency, and the models generated using this approach exhibit strong performance. The resulting model, optimized for hyperparameters, was applied to predict the training data.
The analysis revealed that the alkali content in the washing solution had the greatest influence on the degree of soil cleaning, while other parameters were statistically insignificant. Additionally, no multicollinearity was detected between the factors analyzed. The optimal process parameters for achieving the highest degree of soil purification in the Balakhany region were established: a washing solution containing 0.065 % NaOH, 0.1 % surfactant, 0.5 % solvent, and mechanical agitation for 20 minutes.
As a result of process optimization, the degree of soil purification in the Balakhany region increased by 3 %, reaching 98.9 %. The determination coefficient (R²) for the prediction was calculated as 0.99013, indicating high accuracy. These findings can facilitate the large-scale implementation of the process for cleaning oil-contaminated soils.
References:
1. Babayev M.R. Absheron yarymadasynda neftle chirklenmish torpaglaryn agromelorativ usullarla berpasy (rekultivasiyasy) Azerbaijan Respublikasy Dovlet Neft Shirketi Ekologiya Idaresi “Chirklenmish torpaglarin berpasy-teklif olunan usullar” movzusunda elmi-praktiki konfrans. Meqaleler toplusu, 2008, s. 34-40.
2. Geribov G.S., Khalilov M.U., Memmedova S.Z. Agroekologiya. – Baky: Elm, 2010, 525 s.
3. Gasanov K.S., Abdullaev F.Z. Vybor skhemy pererabotki nefteshlamov i neftezagryaznennykh gruntov, opytno-promyshlennoe ispytanie tekhnologii ikh utilizacii i obezvrezhivaniya // Zhurnal “khimicheskiye problemy“, 2009, № 2,
s. 213-227.
4. Hasanov G., Abdulaev F., Latifov I. Theoretical substantiation of process of rectification of oil-polluted soil. International conference on Environment: Survival and Sustainability. 19-25 February 2007, Nicosia - Northern Cyprus, pp. 3433-3442
5. Gasanov K.S., Abdullaev F.Z., Lyatifov F.I. Ostatochnaya neft’ neftezagryaznennoy pochvy poluostrova Absheron: Obrazovanie, izvlechenie iz pochvy, racional’noe ispol’zovanie // Processy neftekhimii i neftepererabotki, 2006, № 2 (25), s. 62-70.
6. Gasanov K.S., Abdullaev F.Z. Issledovanie poverkhnostnoy aktivnosti ostatochnykh neftey, vydelennykh iz neftezagryaznennoy pochvy poluostrova Absheron // Zhurnal “khimicheskiye problemy“, 2008 № 4, s. 615-623.
7. Hasanov G.S., Valiyev V.G., Aliyeva N.F., Aliyev E.A. Investigation of purification process of oil-contaminated soils. Chemical problems, 2019, № 4 (17), рp. 626-632.
8. Features of the influence of mechanical and chemical impact on the treatment process of oil contaminated soils. Reliabiliti: Theory and Applications, November 2022, special issue volume № 4 (70), 17, pp. 309-317.
9. Gasanov K.S., Abdullaev F.Z., Annagiev M.Kh. Issledovanie form koagulyacionnogo strukturogo brazovaniya sedimentacionnogo neftezagryaznennogo osadka i ocenka stepeni razrusheniya struktury // Zhurnal “khimicheskiye problemy“, 2008, № 3, s. 438-447.
10. Gasanov K.S., Abdullaev F.Z., Annagiev M.KH. Issledovaniya form svyazi vody i immobilizaciya nefti v neftezagryaznennoy pochve // Protsessy neftekhimii i neftepererabotki, 2008, 3-4(35-36), s. 279-288.
11. Dreyper N., Smit G. Prikladnoy regressionnyy analiz, 3-e izdanie, 2016, 912 s.
12. Renner A.G., Aralbaeva G.G., Zinov’eva O.A. Korrelyacionno-regressionnyy analiz. Metodicheskie ukazaniya k laboratornomu praktikumu. -Orenburg: GOU OGU, 2002, 26 s.
13. Aliev R.A., Fazlollahi B., Aliev R.R., Guirimov B.G. Linguistic time series forecasting using fuzzy recurrent neural network. J. Knowledge. “Education” Society of Azerbaijan Republic. Series Business: 3–13, 2005.
14. Aliyarov R.Y., Gardashova L.A., Hasanli N.I. Predicting porosity through fuzzy logic based methods from south caspian basin data. In: Aliev R.A., Kacprzyk J., Pedrycz W., Jamshidi M., Babanli M., Sadikoglu F.M. (eds.) ICAFS 2020. AISC, vol. 1306, pp. 268-274. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-64058-3_33
15. Ercan O. Artificial neural network based modelling of the Marshall Stability of asphalt concrete. Expert Syst. Appl. 38(5), 6025–6030,2011. https://doi.org/10.1016/j.eswa.2010. 11.018
16. Aliev R.A., Gardashova L.A. Z-set based approach to control system design. In: Aliev R.A., Kacprzyk J., Pedrycz W., Jamshidi M., Babanli M., Sadikoglu F.M. (eds.) ICAFS2020. AISC, vol. 1306, pp. 10-21. Springer, Cham 2021. https://doi.org/10.1007/978-3-030-640 58-3_2
17. Gardashova L.A. Z-Set based inference using ALI-2 implication for control system design. In: Aliev R.A., Kacprzyk J., Pedrycz W., Jamshidi M., Babanli M., Sadikoglu F.M. (eds.) ICSCCW 2021. LNNS, vol. 362, pp. 75-84. Springer, Cham 2022. https://doi.org/10.1007/ 978-3-030-92127-9_14
18. Aliyeva K.R. Demand forecasting for manufacturing under Z information. Procedia Comput.Sci.120,5095142017.https://doi.org/10.1016/j.procs.2017.11.272
