For secondary and tertiary oil recovery processes, the interfacial interactions between injected gas and crude oil contributes significantly at the pore-scale. We experimentally examine the effect of ...temperatures, pressures and surfactants on interfacial interaction, contact angle and oil displacement efficiency of CO2 flooding at the pore-scale using microfluidics. Two types of flow channel designs are utilized in this study, blind end of direct channels and curved channels with different diameters. The results indicate that increasing pressure and adding surfactants can promote the contact between oil and gas and reduce the interfacial tension, which has a certain impact on oil displacement efficiency. In the process of gas flooding, CO2 dissolution, expansion and miscibility are the key oil-gas interactions. The results of experiments in curved and direct channels show that the curved channels increase the resistance of fluid flow and limit the contact between CO2 and oil. And with the decrease of the porous medium radius, the capillary action in the pores increases, and the higher the viscous force needed for displacing the oil, which will affect the interaction between oil and gas and the oil recovery in micro pores. Then, two non-ionic alkoxylated surfactants (i.e. ethylene glycol butyl ether and Span 80) were selected to enhance the interaction between CO2 and crude oil. The presence of surfactants increases the solubility of CO2 in crude oil, which reduces the interfacial tension of the system and results in higher oil recovery.
•A series of microscopic experiments were performed to investigate the interaction between CO2 and crude oil by a simple and visual way, and the effect of surfactants on oil recovery was directly reflected.•Different diameters of direct channels and curved channels were used to simulate complex porous media.•The influence of porous media on oil - gas miscibility was studied by analyzing the interaction between oil and gas in the blind end of direct channels and curved channels with different diameters. The effect of temperature and pressure on the mechanisms of oil-gas interaction and supercritical CO2 -surfactant flooding was also investigated.•Based on image processing, a quantitative calculation of contact angle and recovery was developed to assess the effect of miscibility and oil displacement.•Two non-ionic alkoxylated surfactants (i.e. ethylene glycol butyl ether and Span 80) were selected to enhance the interaction between CO2 and crude oil. In the case of the existence of surfactants, stronger oil-gas interactions and higher oil recovery were observed.
Deep learning recommendation models (DLRMs) are used across many business-critical services at Facebook and are the single largest AI application in terms of infrastructure demand in its ...data-centers. In this paper we discuss the SW/HW co-designed solution for high-performance distributed training of large-scale DLRMs. We introduce a high-performance scalable software stack based on PyTorch and pair it with the new evolution of Zion platform, namely ZionEX. We demonstrate the capability to train very large DLRMs with up to 12 Trillion parameters and show that we can attain 40X speedup in terms of time to solution over previous systems. We achieve this by (i) designing the ZionEX platform with dedicated scale-out network, provisioned with high bandwidth, optimal topology and efficient transport (ii) implementing an optimized PyTorch-based training stack supporting both model and data parallelism (iii) developing sharding algorithms capable of hierarchical partitioning of the embedding tables along row, column dimensions and load balancing them across multiple workers; (iv) adding high-performance core operators while retaining flexibility to support optimizers with fully deterministic updates (v) leveraging reduced precision communications, multi-level memory hierarchy (HBM+DDR+SSD) and pipelining. Furthermore, we develop and briefly comment on distributed data ingestion and other supporting services that are required for the robust and efficient end-to-end training in production environments.