our four stage process
Our typical "CFD as a Service" will have four stages.
Pre-Processing, Processing, Post Processing or Results Analysis and Validation/Verification The Cofan "CFD as a Service" process begins with a initial consultation or coversation with the client. During this Requirements stage of the process our client provides and in-depth technical insight into their products and defines their goals and objectives as part of an in-depth discussion between the respective engineering teams. It is during this phase that the Problem definition clearly defines the physical problem you want to study, including the geometry, boundary conditions, fluid properties, and other relevant parameters.
Pre Processing or Problem Definition Requirements
Typical Questions that will help to define the parameters of the simulation in a accurate way is explored during the Pre-Processing stage of the consultation process are:
- The objectives of the analysis
- Easiest way to achieve those objectives
- Which Geometry(ies) should be included
- Freestream and/or environmental operating conditions
- Dimensionality of the spatial model is required (1D)
- What should the flow domain look like
- What temporal modeling is appropriate
- Nature of the viscous flow (convection, laminar, turbulent)
- Thermal limits of components (junction or case temperature)
- Thermal design inputs required by the customer from Cofan
- Range of ambient temperatuires for simulation
- What types of analysis (component, board, system)
- Will heatsink optimization / design services required
- Will physical / empirical testing be required in support of simulations
Required Input from the Customer
- CAD Models of all Components, Hardware and Systems
- Power Levels of all components
- Fan Performances Curves
- Thermal Performance Objectives
- Initial and Environmental Boundary Conditions
Required Output from the CFD Study
- Definition of project scope and deliverables
- Phased timeline
- List of Deliverables
- Junction or Case Temperature of specified Components
- Fluid flow conditions documentation
- Reports, Design/engineering recommendations, databases, etc
- Cost Breakdown
Processing or Solver Phase
The processing phase can be computationally intensive, and the required time for simulation varies depending on factors such as the complexity of the geometry, the desired accuracy, the chosen numerical methods, and the computing resources available. Efficient processing often involves a balance between accuracy and computational cost, as well as careful consideration of solver settings and convergence criteria.
The processing or solver phase is divided up into multiple steps
"Virtual" computational mesh established that govern computational equations and how they are applied using numerical methods, such as finite difference, finite volume, or finite element methods.
The initial set of conditions for the fluid flow, temperature distribution, pressure, and other relevant parameters within the computational domain, to mimic real-world starting state(s) of the system.
Time Stepping (if applicable)
Simulations can progress in "time steps", to better identify problems. At each time step, the solver re-calculates the flow and related properties over the computational mesh. The time step size is determined by factors like stability and temporal accuracy desired.
Repeated value updates of flow properties (velocity, pressure, temperature, etc.) across the entire computational mesh. The solver iterates until a certain convergence criterion is met, indicating a stable state or acceptable level of accuracy.
Boundary conditions are applied to the computational domain to reflect interactions between the system and its surroundings. Conditions include inflow, outflow, wall conditions, symmetry conditions, and more.
Post Processing Output or Analysis
Contour plots display variations of a specific parameter (e.g., velocity, pressure, temperature) across the computational domain. Color-coded contours help visualize gradients and trends.
Vector Plots represent velocity vectors at different locations in the domain, indicating flow directions and magnitudes.
Streamlines depict the path that fluid particles follow within the flow field, providing a clear visualization of flow patterns.
Showing the trajectories of individual fluid particles released at different points over time, helping to understand flow behavior over longer periods.
Surface Pressure Distribution: This plot shows the pressure distribution on the surfaces of objects within the domain, helping identify regions of high and low pressure. Heat Transfer Analysis: Surface temperature distribution plots provide insights into the heat transfer characteristics of the system.
These representations involve cutting through the computational domain to visualize internal flow structures, temperature gradients, or other properties along specific planes or sections.
For transient simulations, animations can be created to visualize the evolution of flow patterns over time. Time-averaged data helps understand statistical properties of the flow, such as mean velocity profiles or temperature distributions.
Extracting quantitative data, such as maximum/minimum values, average values, or integrated quantities (e.g., mass flow rates), for specific regions of interest within the domain.
Generating reports that summarize key simulation results and findings.
Conclusions about flow behavior, identifying areas of interest, and making design decisions based on the analysis of simulation results and able to be drawn from concatenating the different types of plots
Validation and Verification
Data for simulation validation could come from any number of physical measurements in a controlled environment, such as wind tunnel tests, flow visualization techniques, thermocouple measurements or other relevant sources. Overlaying the simulation results with experimental data provides one methodology to validate the accuracy of the simulation.
It's important to note that perfect agreement between simulation and experimental data might not always be achievable due to uncertainties in both the simulation setup and experimental measurements. However, a reasonable level of agreement within acceptable tolerances demonstrates that the simulation model is reliable for its intended application.