Department of Chemical Engineering, University of Zanjan, Zanjan, Iran.
Chemical Engineering Department, Urmia University, Urmia, Iran.
Environ Sci Pollut Res Int. 2022 Feb;29(10):13767-13781. doi: 10.1007/s11356-021-16568-6. Epub 2021 Oct 1.
To commercialize the biocementation through microbial induced carbonate precipitation (MICP), the current study aimed at replacing the costly standard nutrient medium with corn steep liquor (CSL), an inexpensive bio-industrial by-product, on the production of urease enzyme by Sporosarcina pasteurii (PTC 1845). Multiple linear regression (MLR) in linear and quadratic forms, adaptive neuro-fuzzy inference system (ANFIS), and genetic programming (GP) were used for modeling of process based on the experimental data for improving the urease activity (UA). In these models, CSL concentration, urea concentration, nickel supplementation, and incubation time as independent variables and UA as target function were considered. The results of modeling showed that the GP model had the best performance to predict the extent of urease, compared to other ones. The GP model had higher R as well as lower RSME in comparison with the models derived from ANFIS and MLR. Under the optimum conditions optimized by GP method, the maximum UA value of 3.6 Mm min was also obtained for 5%v/v CSL concentration, 4.5 g L urea concentration, 0 μM nickel supplementation, and 60 h incubation time. A good agreement between the outputs of GP model for the optimal UA and experimental result was obtained. Finally, a series of laboratory experiments were undertaken to evaluate the influence of biological cementation on the strengthening behavior of treated soil. The maximum shear stress improvement between bio-treated and untreated samples was 292% under normal stress of 55.5 kN as a result of an increase in interparticle cohesion parameters.
为了通过微生物诱导碳酸钙沉淀（MICP）将生物胶结商业化，本研究旨在用廉价的生物工业副产物玉米浆（CSL）代替昂贵的标准营养培养基，用于生产 Sporosarcina pasteurii（PTC 1845）的脲酶。基于实验数据，采用多元线性回归（MLR）的线性和二次形式、自适应神经模糊推理系统（ANFIS）和遗传编程（GP）来建模，以提高脲酶活性（UA）。在这些模型中，CSL 浓度、尿素浓度、镍补充和孵育时间作为自变量，UA 作为目标函数。建模结果表明，与其他模型相比，GP 模型在预测脲酶程度方面表现最佳。与源自 ANFIS 和 MLR 的模型相比，GP 模型具有更高的 R 和更低的 RSME。在 GP 方法优化的最佳条件下，还获得了 5%v/v CSL 浓度、4.5 g/L 尿素浓度、0 μM 镍补充和 60 h 孵育时间下最大 UA 值为 3.6 Mm min。GP 模型对最佳 UA 的输出与实验结果之间具有良好的一致性。最后，进行了一系列实验室实验来评估生物胶结对处理土壤强化行为的影响。在 55.5 kN 的正常应力下，生物处理和未处理样品之间的最大剪切应力提高了 292%，这是由于颗粒间内聚力参数的增加。
