arXiv subjects

Sibendu Som

Publications and source records attributed to Sibendu Som.

4 recordsLinked to original sources

arXiv2025

AISAC: An Integrated multi-agent System for Transparent, Retrieval-Grounded Scientific Assistance

AI Scientific Assistant Core (AISAC) is a transparent, modular multi-agent runtime developed at Argonne National Laboratory to support long-horizon, evidence-grounded scientific reasoning. Rather than proposing new agent algorithms or claiming autonomous scientific discovery, AISAC contributes a governed execution substrate that operationalizes key requirements for deploying agentic AI in scientific practice, including explicit role semantics, budgeted context management, traceable execution, and reproducible interaction with tools and knowledge. AISAC enforces four structural guarantees for scientific reasoning: (1) declarative agent registration with runtime-enforced role semantics and automatic system prompt generation; (2) budgeted orchestration via explicit per-turn context and delegation depth limits; (3) role-aligned memory access across episodic, dialogue, and evidence layers; and (4) trace-driven transparency through persistent execution records and a live event-stream interface. These guarantees are implemented through hybrid persistent memory (SQLite and dual FAISS indices), governed retrieval with agent-scoped RAG, structured tool execution with schema validation, and a configuration-driven bootstrap mechanism that enables project specific extension without modifying the shared core. AISAC is currently deployed across multiple scientific workflows at Argonne, including combustion science, materials research, and energy process safety, demonstrating its use as a reusable substrate for domain-specialized AI scientific assistants.

cs.AI↗

arXiv2018

Exploration of cavitation-induced erosion metrics in throttle flow simulations

Although there have been extensive experimental and computational investigations in characterizing cavitation phenomenon in both diesel and gasoline direct injectors, much is still unknown about the mechanisms driving cavitation-induced erosion, and how this complicated fluid-structure interaction should be modeled. To explore current modeling capabilities, a numerical investigation was conducted within the CONVERGE modeling framework to assess proposed cavitation erosion metrics in the literature, and their link to the predicted cavitation cloud collapse mechanism. The multiphase flow within the Winklhofer Throttle U geometry was modeled using a compressible mixture model, where phase change was represented using the Homogeneous Relaxation Model (HRM) and the turbulent flow evolution was modeled using a dynamic structure approach for Large Eddy Simulations (LES). After comparing the model predictions against available experimental data, representative condensation events and potential cavitation erosion sites were identified. The cavitation cloud structures responsible for potential material damage were visualized through the evolution of the vorticity field. For the modeled throttle geometry, it was found that the horseshoe cloud implosion mechanism was predicted to occur and generate excessive impact loads at the throttle boundary.

physics.flu-dyn↗

arXiv2018

Evaluation of a new cavitation erosion metric based on fluid-solid energy transfer in channel flow simulations

Although there have been extensive investigations characterizing cavitation phenomenon in fuel injectors, much is still unknown about the mechanisms driving cavitation-induced erosion, and how these complicated physics should be represented in a model. In lieu of computationally expensive fluid-structure interaction modeling, the Eulerian mixture modeling approach has been accepted as an efficient means of capturing cavitation phenomena. However, there remains a need to link the erosive potential of cloud collapse events with the subsequent material deformation and damage of neighboring surfaces. Even though several cavitation erosion indices have been proposed in the literature, no single metric has been identified as universally applicable across all injector-relevant conditions. The objective of this work is to identify parameters that characterize the erosive potential of cavitation cloud collapse mechanisms that are likely to occur within injector orifices. While a commonly employed cavitation erosion metric, namely the maximum local pressure, was found to provide indications of potential sites for pitting and material rupture from single impact events, no additional information could be determined regarding the material erosion process. To improve representation of the incubation period within the cavitation erosion process, a new metric was derived based on cumulative energy absorbed by the solid material from repeated hydrodynamic impacts. Through evaluation of predicted cavitation cloud collapse events in a channel geometry against available experimental data, the stored energy metric yielded insight into the erosive potential of recorded impact events. The stored energy metric provided a means to accurately predict the influence of flow conditions on the incubation period before material erosion. ...

physics.flu-dyn↗

arXiv2018

A Numerical Study on Spray Characteristics at Start of Injection for Gasoline Direct Injection

This paper presents a CFD study of Engine Combustion Network (ECN) Spray G, focusing on the transient characteristics of spray at start of injection. The Large Eddy Simulation (LES) coupled with Volume of Fluid (VOF) method is used to model the turbulent two-phase flow. A moving needle boundary condition is applied to capture the internal flow condition accurately. The injector geometry was measured with micron-level resolution using full spectrum x-ray tomographic imaging at Advanced Photon Source (APS) at Argonne National Labora-tory, providing detailed machining error from manufacturer and realistic rough surface. For comparison, a nominal geometry is also used for the simulation. Spray characteristics such as Sauter Mean Diameter (SMD), droplet volume and surface area are extracted by post-processing the CFD outputs. It is seen that compared to the nominal geometry, the use of the high resolution real geometry predicts about 11% lower SMD. The rough surface along with manufacturing defects generate more unstable structures along the jet surface and accelerates the liquid de-formation and breakup processes, starting inside of the counter-bore. This result shows that the machining details of injector, which is usually ignored in the two-phase flow simulations, has considerable impact on the spray development process.

physics.flu-dyn↗