Search

A numerical method based on smoothed particle hydrodynamics with adaptive spatial resolution (SPH-ASR) was developed for simulating free surface flows. This method can reduce the computational demands while maintaining the numerical accuracy. In this method, the spatial resolution changes adaptively according to the distance to the free surface by numerical particle splitting and merging. The particles are split for refinement when they are near the free surface, while they are merged for coarsening when they are away from the free surface. A search algorithm was implemented for identifying the particles at the free surface. A particle shifting technique, considering variable smoothing length, was introduced to improve the particle distribution. The presented SPH-ASR method was validated by simulating various free surface flows, and the results were compared to those obtained using SPH with uniform spatial resolution (USR) and experimental data., Comment: 16 figures

It is common practice to use large computational resources to train neural networks, as is known from many examples, such as reinforcement learning applications. However, while massively parallel computing is often used for training models, it is rarely used for searching solutions for combinatorial optimization problems. In this paper, we propose a novel massively parallel Monte-Carlo Tree Search (MP-MCTS) algorithm that works efficiently for 1,000 worker scale, and apply it to molecular design. This is the first work that applies distributed MCTS to a real-world and non-game problem. Existing work on large-scale parallel MCTS show efficient scalability in terms of the number of rollouts up to 100 workers, but suffer from the degradation in the quality of the solutions. MP-MCTS maintains the search quality at larger scale, and by running MP-MCTS on 256 CPU cores for only 10 minutes, we obtained candidate molecules having similar score to non-parallel MCTS running for 42 hours. Moreover, our results based on parallel MCTS (combined with a simple RNN model) significantly outperforms existing state-of-the-art work. Our method is generic and is expected to speed up other applications of MCTS.

We introduce Bee$^+$, a 95-mg four-winged microrobot with improved controllability and open-loop-response characteristics with respect to those exhibited by state-of-the-art two-winged microrobots with the same size and similar weight (i.e., the 75-mg Harvard RoboBee). The key innovation that made possible the development of Bee$^+$ is the introduction of an extremely light (28-mg) pair of twinned unimorph actuators, which enabled the design of a new microrobotic mechanism that flaps four wings independently. A first main advantage of the proposed design, compared to those of two-winged flyers, is that by increasing the number of actuators from two to four, the number of direct control inputs increases from three to four when simple sinusoidal excitations are employed. A second advantage of Bee$^+$ is that its four-wing configuration and flapping mode naturally damp the rotational disturbances that commonly affect the yaw degree of freedom of two-winged microrobots. In addition, the proposed design greatly reduces the complexity of the associated fabrication process compared to those of other microrobots, as the unimorph actuators are fairly easy to build. Lastly, we hypothesize that given the relatively low wing-loading affecting their flapping mechanisms, the life expectancy of Bee$^+$s must be considerably higher than those of the two-winged counterparts. The functionality and basic capabilities of the robot are demonstrated through a set of simple control experiments., Comment: Accepted for publications in IEEE Robotics and Automation Letters (RA-L) and the Proceedings of the 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2019). 8 pages, 6 figures

The smoothed particle hydrodynamics (SPH) method has been increasingly used to study fluid problems in recent years; but its computational cost can be high if high resolution is required. In this study, an adaptive resolution method based on SPH is developed for multiphase flow simulation. The numerical SPH particles are refined or coarsened as needed, depending on the distance to the interface. In developing the criteria, reference particle spacing is defined for each particle, and it changes dynamically with the location of the interface. A variable smoothing length is used together with adaptive resolution. An improved algorithm for calculating the variable smoothing length is further developed to reduce numerical errors. The proposed adaptive resolution method is validated by five examples involving liquid drops impact on dry or wet surfaces, water entry of a cylinder and dam break flow, with the consideration of ambient gas. Different resolution levels are used in the simulations. Numerical validations have proven that the present adaptive resolution method can accurately capture the dynamics of liquid-gas interface with low computational costs. The present adaptive method can be incorporated into other SPH-based methods for efficient fluid dynamics simulation., Comment: 23 figures

Natural convection is of great importance in many engineering applications. This paper presents a smoothed particle hydrodynamics (SPH) method for natural convection. The conservation equations of mass, momentum and energy of fluid are discretized into SPH equations. The body force due to the change of density in a temperature field is considered by the Boussinesq approximation. The numerical method is validated by comparing numerical results with experimental results from literature. The numerical and experimental results reach a good agreement. Then the SPH method is applied to study the natural convection in a horizontal concentric cylindrical annulus with Rayleigh number in the range of 10^2 to 10^7 and Prandtl number in the range of 0.01 to 10. In general, the flow is stable at low Rayleigh number but unstable at high Rayleigh number. The transition Rayleigh number from stable to unstable states is lower in the low Prandtl number cases than in the high Prandtl number cases. Four different convection states are identified in numerical simulations, namely, stable state with 1 plume (SP1), unstable state with 1 plume (UP1), stable state with n (n > 1) plumes (SPN), and unstable state with n (n > 1) plumes (UPN). The SP1 and UP1 states are observed for all Prandtl numbers, while the SPN and UPN states are only observed at Pr = 0.1 and 0.01., Comment: 7 figures

The interaction of liquid drops and heated surfaces is of great importance in many applications. This paper describes a numerical method, based on smoothed particle hydrodynamics (SPH), for simulating n-heptane drop impact on a heated surface. The SPH method uses numerical Lagrangian particles, which obey the laws of fluid dynamics, to describe the fluid flows. By incorporating the Peng-Robinson equation of state, the present SPH method can directly simulate both the liquid and vapor phases and the phase change process between them. The numerical method was validated by two experiments on drop impact on heated surfaces at low impact velocities. The numerical method was then used to predict drop-wall interactions at various temperatures and velocities. The model was able to predict the different outcomes, such as rebound, spread, splash, breakup, and the Leidenfrost phenomenon, consistent with the physical understanding., Comment: 8 figures

The impact of liquid drops on a heated solid surface is of great importance in many engineering applications. This paper describes the simulation of the drop-wall interaction using the smoothed particle hydrodynamics (SPH) method. The SPH method is a Lagrangian mesh-free method that can be used to solve the fluid equations. A vaporization model based on the SPH formulation was also developed and implemented. A parametric study was conducted to characterize the effects of impact velocity and wall temperature on the impact outcome. The present numerical method was able to predict different outcomes, such as deposition, splash, breakup, and rebound (i.e., Leidenfrost phenomenon). The present numerical method was used to construct a regime diagram for describing the impact of an iso-octane drop on a heated surface at various Weber numbers and wall temperatures.

Automatic design of organic materials requires black-box optimization in a vast chemical space. In conventional molecular design algorithms, a molecule is built as a combination of predetermined fragments. Recently, deep neural network models such as variational auto encoders (VAEs) and recurrent neural networks (RNNs) are shown to be effective in de novo design of molecules without any predetermined fragments. This paper presents a novel python library ChemTS that explores the chemical space by combining Monte Carlo tree search (MCTS) and an RNN. In a benchmarking problem of optimizing the octanol-water partition coefficient and synthesizability, our algorithm showed superior efficiency in finding high-scoring molecules. ChemTS is available at https://github.com/tsudalab/ChemTS.

This study presents a smoothed particle hydrodynamics (SPH) method with Peng-Robinson equation of state for simulating drop vaporization and drop impact on a hot surface. The conservation equations of momentum and energy and Peng-Robinson equation of state are applied to describe both the liquid and gas phases. The governing equations are solved numerically by the SPH method. The phase change between the liquid and gas phases are simulated directly without using any phase change models. The numerical method is validated by comparing numerical results with analytical solutions for the vaporization of n-heptane drops at different temperatures. Using the SPH method, the processes of n-heptane drops impacting on a solid wall with different temperatures are studied numerically. The results show that the size of the film formed by drop impact decreases when temperature increases. When the temperature is high enough, the drop will rebound., Comment: 6 pages, 4 figures, 10th U.S. National Combustion Meeting, April 23-26, 2017, College Park, Maryland

Smoothed particle hydrodynamics (SPH) method has been increasingly used for simulating fluid flows, however its ability to simulate evaporating flow requires significant improvements. This paper proposes an SPH method for evaporating multiphase flows. The present SPH method can simulate the heat and mass transfers across the liquid-gas interfaces. The conservation equations of mass, momentum and energy were reformulated based on SPH, then were used to govern the fluid flow and heat transfer in both the liquid and gas phases. The continuity equation of the vapor species was employed to simulate the vapor mass fraction in the gas phase. The vapor mass fraction at the interface was predicted by the Clausius-Clapeyron correlation. A new evaporation rate was derived to predict the mass transfer from the liquid phase to the gas phase at the interface. Because of the mass transfer across the liquid-gas interface, the mass of an SPH particle was allowed to change. New particle splitting and merging techniques were developed to avoid large mass difference between SPH particles of the same phase. The proposed method was tested by simulating three problems, including the Stefan problem, evaporation of a static drop, and evaporation of a drop impacting on a hot surface. For the Stefan problem, the SPH results of the evaporation rate at the interface agreed well with the analytical solution. For drop evaporation, the SPH result was compared with the result predicted by a level-set method from literature. In the case of drop impact on a hot surface, the evolution of the shape of the drop, temperature, and vapor mass fraction were predicted., Comment: 26 pages, 15 figures