AICFD — Artificial Intelligence Computational Fluid Dynamics


AICFD is universal computational fluid dynamics (CFD) simulation software that is independently developed by Nanjing Tianfu Software Co., Ltd. to quickly and intelligently simulate flows and heat transfer. AICFD has several features, including model import, quick and automatic mesh generation, quick simulation, result visualization and post-processing, and intelligent acceleration. These features empower the entire simulation analysis process from building geometric models to analyzing simulation results. Relying on a modern GUI and numerical simulation and intelligent acceleration algorithms, AICFD provides users with easy-to-use features for intelligent CFD simulation. AICFD also helps industrial enterprises establish an integrated design, simulation, and optimization process, thereby greatly improving their product development efficiency.


(1) One-click simulation

Most commercial simulation software products on the market are specially designed for simulation engineers. These software products require complex operations and high learning costs and are less user-friendly to design engineers. AICFD provides a graphical and integrated simulation process in which users need only to set the required parameters. AICFD automatically proceeds with subsequent complex operations, such as mesh generation, computing, and post-processing. This process makes AICFD more user-friendly to design engineers.


Figure 1 One-click simulation

(2) Industrial design-oriented CFD simulation features 

AICFD provides common CFD simulation features in industrial design. The supported flow types include single-phase incompressible flows, single-phase compressible flows (including subsonic, transonic, and supersonic flows), heat transfer, and multiphase flows. AICFD can simulate flows and heat transfer across regions, making it capable of simulating complex industrial flows, such as flows and heat transfer in turbomachinery and heat exchangers. AICFD provides various robust numerical schemes and boundary conditions and common physical models. Its universal CFD simulation features appeal to design engineers in various fields, such as resources and power, shipbuilding and marine engineering, aerospace, and automobile.


Figure 2 Abundant CFD simulation features

(3) Quick and intelligent simulation and real-time simulation

Other commercial simulation software products on the market take several hours, days, or even weeks to complete a simulation task. AICFD uses AI technologies to accelerate simulation, thereby shortening the simulation time to seconds and significantly improving the simulation efficiency. AICFD deeply integrates simulation and AI technologies to implement real-time simulation for specific models. The quick and intelligent simulation and real-time simulation features of AICFD make it a good choice for design engineers in routine simulation work.


Figure 3 Quick and intelligent simulation and real-time simulation

(4) Versatility and scalability

As universal CFD simulation software, AICFD can be used in a wide range of fields due to the versatility of its core computing module. AICFD thoroughly analyzes simulation processes for dedicated simulation fields to optimize the simulation module and improve simulation accuracy and operational convenience.

(5) User-friendly GUI and adaptation to multiple operating systems

AICFD provides a client-based GUI that can meet the requirements of complex and heavyweight simulation. Both Windows and Linux versions of AICFD are available.


(1) Cyclone separator

The cyclone separator in this case is a common type of separation and classification device that removes particles from fluid based on the principle of centrifugal sedimentation. The flow in this case is incompressible. The fluid is fed into the cyclone separator through the inlet and rotates downwards to the bottom of the cyclone separator. Then, the fluid rotates upwards and is discharged through the outlet.


Figure 4 Flow pattern of fluid in the cyclone separator (velocities are represented by different colors)

(2) Flow around the ONERA M6 wing

The M6 wing in this case is a wing model designed by the French Aerospace Lab Office, Office National d'Etudes et de Recherches Aérospatiales (ONERA). This model has undergone a series of transonic wind-tunnel tests and is supported by abundant experimental data. Although the M6 wing uses simple geometry, the transonic flow around it is complex, involving at least the local supersonic flow, shock waves, and boundary layer separation. The flow around the M6 wing has the typical features of a three-dimensional (3D) compressible flow. Lambda-shaped shock waves are generated on the wing surface. Therefore, the flow around the M6 wing is usually used as a verification case for CFD software.


Figure 5 Distribution of pressure on the wing surface caused by the flow around the M6 wing

(3) Turbomachinery

The centrifugal pump in this case works at a rotation speed of 1,770 revolutions per minute (RPM) with water as the working medium. Figure 6 shows the geometry of the centrifugal pump.


Figure 6 Geometry of the centrifugal pump after meshing


 Figure 7 Pressure nephogram of the centrifugal pump

(4) Automobile

This case computes the external flow field around a DrivAer passenger car model and provides a large amount of experimental data for comparison. As shown in figure 8, a three-row passenger car model with a fastback and a smooth underbody is used in this case. The influx velocity is 30 m/s. The ground is modeled as a moving wall with a velocity of 30 m/s. The wheels are assigned a rotating wall boundary condition, and the rotation speed is 94 rad/s.


Figure 8 DrivAer passenger car model with a fastback and a smooth underbody


Figure 9 Pressure nephogram of the DrivAer passenger car model

Table 1 Comparison between the calculated values of AICFD and experimental values



1. Experimental Comparison of the Aerodynamic Behavior of Fastback and Notchback DrivAer Models. SAE Technical Paper 2014-01-0613, 2014

(5) Shipbuilding

In this case, the Volume of Fluid (VOF) method of AICFD is used to perform numerical simulation for the external flow field of a Wigley hull. This case provides a large amount of experimental data for comparison. In this case, the k-omega shear-stress transport (SST) turbulence model with a wall function is used. The range of y+ is from 30 to 150, and the Froude number is 0.267.


Figure 10 Mesh of the Wigley hull


Figure 11 Hull on the water 

(6) Real-time simulation

This case analyzes the external flow field of a 3D Ahmed body. The following tables compare the simulation results between AICFD and another CFD software product in the condition that the slant angle at the rear end of the Ahmed body is 25.2°.


Figure 12 Mesh distribution of the Ahmed body

Table 2 Comparison of the simulation results for the moment coefficient, drag coefficient, and lift coefficient between AICFD and another CFD software product


Table 3 Comparison of the simulation results for the fields in the central section between AICFD and another CFD software product


This case uses 198,633 meshes. In the condition of four-core parallel computing for 1,000 steps and a steady-state residual of 1e-4, the other CFD software product takes 400s to complete computation, whereas AICFD takes only 0.036s. This indicates that AICFD can be used to perform real-time simulation.