4.3.2. Nitrobenzene Removal in 2-D Layered Heterogeneous Porous Media

When R = 17.4 and steam injection time was 3 h, the NB concentration in the center of the 2-D simulation tank (the position closest to the steam injection point) was the lowest, and the concentration on both sides was high (Figure 7a). This was primarily due to the effect of buoyancy. The vertical migration of steam was faster, resulting in the preferential removal of vertical pollutants. After 6 h of steam injection, the concentration distribution of NB was affected by the interface, and the concentration was the lowest below the interface, with a concentration of 60 mg/L. After 6 h, the NB removal rate in the simulated tank was 68%, of which the removal rates in gravel and medium sand were 70.3% and 65.3%, respectively, and the average NB concentration in medium sand was 1.13 times that of gravel.

When R = 381, the blocking effect of the interface was enhanced (Figure 7b). The NB concentration in the upper layer was evidently higher than that in the lower layer after 3 h of steam injection, and the difference was more obvious with the increase in steam injection time. After 6 h of steam injection, the lowest concentration of NB in gravel reached 28 mg/L, and the average concentration of NB in fine sand was 3.13 times that of gravel. The total removal rate of NB was 60.6%, among which the removal rates in gravel and fine sand were 79.9% and 37%, respectively. Compared with R = 17.4, the total removal rate of NB decreased, while the removal rate of NB in gravel increased.

**Figure 6.** Temperature distribution of the layered heterogeneous aquifers (**a**) R = 17.4 (**b**) R = 381.

**Figure 7.** Isograms of NB concentrations with different R values: (**a**) R = 17.4; (**b**) R = 381.

4.3.3. Alcohol-Enhanced Steam Remediation of Layered Heterogeneous Aquifer Contaminated by Nitrobenzene

After adding ethanol, some heat was accumulated at the interface, but the temperature difference between the upper and lower layers decreased significantly, the average temperature increased by 15 ◦C, and the area affected by steam in the low-permeability zone increased by 35% (Figure 8). Alexeev et al. (2005) found that Marangoni convection could enhance liquid phase heat transfer through experimental and numerical simulation [41].

Therefore, the addition of ethanol produces Marangoni convection and enhances mass and heat transfer.

**Figure 8.** Temperature distribution without ethanol (**a**) and with ethanol (**b**) at R = 381.

After adding ethanol, the remediation effect of NB was significantly increased, particularly in the fine sand medium (Figure 9) and the average NB concentration in the simulated tank was reduced by 51%. The average concentration of NB in the lower gravel and upper fine sand decreased by 43.7% and 43.5%, respectively, indicating that ethanol addition enhanced the gas–liquid mass transfer. During the repair process, ethanol volatilized at the gas–liquid interface, forming a surface tension gradient at the interface and increasing the convective vortices in the flow field, resulting in the Marangoni convection [20,40,42]. The Marangoni convection can also enhance heat transfer, and an increase in temperature also enhances NB removal. In addition, ethanol reduces the azeotropic temperature of NB, promoting the boiling of NB at 83.2 ◦C, thus, improving the removal rate and efficiency of NB.

4.3.4. Alcohol-Enhanced Steam Remediation of Lens Heterogeneous Aquifer Contaminated with NB

Figure 10 shows the temperature distribution of the aquifers containing lenses with and without ethanol solution. Due to the existence of a low-permeability lens, a blocking interface was formed below the lens, and steam flowed around the lens. Only a small amount of steam could enter the fine-sand lens, causing the temperature inside the lens to increase. However, most steam will flow around the lens from the high-permeability area in the form of preferential flow, resulting in an uneven distribution of steam, which affects the removal effect of NB. After adding ethanol, the average temperature in the simulated tank increased by 8 ◦C, and the steam-affected area increased by 14%, which further confirms that the Marangoni convection generated by ethanol can enhance heat and mass transfer.

**Figure 9.** Isograms of NB concentration without ethanol (**a**) and with ethanol (**b**) at 9 h.

**Figure 10.** Temperature distribution without ethanol (**a**) and with ethanol (**b**) in the lens heterogeneous aquifer.

Figure 11 shows the isolines of NB concentration in the heterogeneous aquifer with the lens. In the lens heterogeneous aquifer, the remediation effect of steam at the edge of the simulation tank and in the lens was weakened due to the preferential flow. After ethanol was added, the mass transfer and heat transfer were enhanced, and the boiling area increased, resulting in an increase in the removal efficiency of NB. The removal efficiency of the tank was increased by 10%, and the NB concentration in the lens was reduced by approximately 58%. Under the same steam flow rate, ethanol addition enhanced the remediation effect of the lens heterogeneous aquifer and reduced the remediation time and cost.

**Figure 11.** Isograms of NB concentration without ethanol (**a**) and with ethanol (**b**) in the heterogeneous aquifer.
