Development of a New Numerical Model to Simulate Chemotaxisdriven Bacterial Transport for Treatment of Tumor Cells and Mitigation of Bacterially-mediated Pipeline Corrosion Problems

The objective of this project was to develop a new, computationally efficient, multiscale numerical model to simulate the directional and tumble motions of self-propelled chemotactic particles (live bacteria, or bacterial or chemical robots) in spatially and temporarily varying chemoattractant (e.g., nutrients) gradients in geometrically complex domains by accommodating cells scale adaptation dynamics and signal processing, and particle-scale fluid-particle hydrodynamics. The project objectives also involved developing new modules to simulate non-Newtonian fluid flows, arbitrary-shape particles flows, and advective-diffusive-reactive transport of substrates to extend the model use for targeted applications in biomedical and oil and gas fields. Prior to this project, such a comprehensive numerical model with these capabilities did not exist. The project involved experimental tasks to study chemotactic motility of an E. coli bacterial strain, with the intention that the experimental data would produce good quality data for the validation of the proposed numerical model.

Major components of the new numerical model were developed based on the LatticeBoltzmann (LB). An existing RapidCell (RC) model was coupled with our-in-house colloidal LB model to simulate chemotactic motility of chemotactic particles. The coupling involved a new set of equations SwRI derived to calculate the position of receptor clusters on the chemotactic particle surfaces through which chemotactic particles would detect and orient toward the transient chemoattractant gradients. The coupling also involved calculations of forces and torques on chemotactic particles due to their run and tumble motions as part of their adaption dynamics The non-Newtonian flow module was developed by relating the local fluid viscosity to the second invariant of the rate-of-strain tensor. The arbitrary-shape particles flow module was developed through new geometrical relations to locate surface boundary nodes via the winding number algorithm of any arbitrary-shaped particles' surfaces along which the particle and the fluid exchange momentum. The advective-diffusive substrate transport module was formulated by relating the relaxation parameter to the molecular diffusion of the substrate through the Chapman Enskog expansion. SwRI impended a series of optimization techniques, including optimizing array memory layouts, data structure simplification, random number generation outside the simulation thread(s), code parallelization via OpenMP, and intra- and inter-timestep task pipelining to enhance the computational performance of the new model. Microfluidic experiments were set up to generate data for model validation.

SwRI successfully developed a new multi-scale model to simulate motility of chemotactic particles in complex fluidic environments. SwRI also successfully formulated new modules for non-Newtonian fluid flows (for pseduoplastic and dilatant fluid flows), advective-diffusive substrate transport, and arbitrary-shaped particles flows (e.g., snapshots from gravity-driven settling of four starshaped particles in an initially quiescent fluid in a bounded domain) and validated these new modules with benchmark problems. Moreover, SwRI successfully implemented optimization techniques to enhance the computational performance of the new model, which resulted in 15 to 40 times computational speed-ups. As for the experimental tasks, although we gained substantial experience in the design and performance of microfluidic experiments to study chemotactic particle motion, the experiments did not produce good quality data to validate the RC-CLB model.