The recovery of levulinic acid (LA) and formic acid (FA) from aqueous solutions and actual biomass hydrolysates by using the stepwise adsorption method was investigated for the first time. The carbon molecular sieve was used as the adsorbent for FA and microporous activated carbon was used as the adsorbent for LA according to the steric hindrance effect of the pore channels and the difference in molecular diameter of FA and LA. The adsorption data of LA and FA onto their respective adsorbents were measured and numerically analyzed by adopting batch adsorption experiment and fixed-bed column methods. In the study of static adsorption equilibrium experiment, the Langmuir equation fitted the data of LA adsorption on activated carbon, while the Freundlich equation fitted well the adsorption data of FA adsorption on carbon molecular sieve. In the kinetic study, the LA adsorption on activated carbon and FA adsorption on carbon molecular sieve confirmed the PSO model. Moreover, the effect of various conditions on fixed-bed column adsorption progress was investigated, which indicated that the adsorption property improved with feed flow rate, column length, temperature decreasing and initial concentration increasing. The Yoon-Nelson model was adopted to predict and analyze the whole process of adsorption breakthrough and consistent well with the experimental data. The separation of LA and FA was realized by continuous fractional adsorption.

首次研究了采用逐步吸附法从水溶液和实际生物质水解物中回收乙酰丙酸（LA）和甲酸（FA）。根据孔道的空间位阻效应以及FA和LA分子直径的差异，采用碳分子筛作为FA的吸附剂，采用微孔活性炭作为LA的吸附剂。采用分批吸附实验和固定床柱法对LA和FA在各自吸附剂上的吸附数据进行了测量和数值分析。在静态吸附平衡实验研究中，Langmuir方程拟合了LA在活性炭上的吸附数据，Freundlich方程很好地拟合了FA吸附在碳分子筛上的吸附数据。在动力学研究中，活性炭上的 LA 吸附和碳分子筛上的 FA 吸附证实了 PSO 模型。此外，研究了各种条件对固定床柱吸附进程的影响，表明吸附性能随着进料流速、柱长、温度降低和初始浓度增加而提高。采用Yoon-Nelson模型对吸附突破全过程进行预测分析，与实验数据吻合较好。LA和FA的分离采用连续分级吸附实现。这表明吸附性能随着进料流速、柱长、温度降低和初始浓度增加而提高。采用Yoon-Nelson模型对吸附突破全过程进行预测分析，与实验数据吻合较好。LA和FA的分离采用连续分级吸附实现。这表明吸附性能随着进料流速、柱长、温度降低和初始浓度增加而提高。采用Yoon-Nelson模型对吸附突破全过程进行预测分析，与实验数据吻合较好。LA和FA的分离采用连续分级吸附实现。