PROJECT REFLECTION

A parameterised workflow was created for the generation of ribbed heat exchangers and built entirely using Open Source software. Parameterised geometry files were created within PyOCC and were imported into OpenFOAM where they were meshed and analysed. Full automation was achieved for this process, with Allrun scripts executing the entire workflow of .stl generation, folder creation, meshing, simulation and post processing from a single run command. For a user specified range of pin height, pin radius, and cell length, any number of possible geometric combinations can be produced.

DISCUSSION OF RESULTS

Varying the pin height from the smallest value to the largest was found to give a decrease in pressure drop due to the fluid having more space to flow freely. A pin height of 5mm proved to have a much larger pressure drop than the other results due to the smaller volume for the fluid to flow through.

Varying the length of the pin distance seemed to have a reasonably small effect on the pressure drop, except for the shortest lengths of 20mm and 30mm, where there is a noticeably larger pressure drop. The close proximity of the pins within the heat exchanger would appear to be the most likely explanation for this, similar to that of reducing the pin height

When the radii of the pins are increased, the pressure change from inlet to outlet increases reasonably linearly until a radius of 22.5mm, before observing an exponential increase beyond that point. Again, as the pin radii increases the flow is subjected to more obstructions, resulting in areas of stagnation and increased velocity gradients elsewhere, combining to result in a larger pressure drop across the cell.

RECOMMENDATIONS

During the project, there were difficulties and improvements that could be made. Recommendations have been made for when this project is carried forward. The recommendations are listed below.

It would be highly recommended that the user becomes proficient in C++ programming, bash scripting and the use of Linux as well as being comfortable using OpenFOAM to a high standard. It would also be advised that the user is comfortable understanding python coding to enable either geometry to be created simply using PyOCC, or even for editing the modelling code to suit different geometries.

It would be recommended that the sequential methodology as conducted in this report is followed as it vastly improves the understanding of the background workings of OpenFOAM.

If only one model is required to be analysed, it is recommended that the user uses ANSYS, if available, instead of OpenFOAM. ANSYS can produce results quickly and efficiently due to its intuitive GUI in all stages of modelling, meshing, simulation and post-processing.

FURTHER WORK

1. Application to Industry

The heat exchanger concept could be applied to many different industries including automotive, electrical and aerospace. The project was initially based on a heat exchanger for electrical components within electric vehicles, however, it is possible that the heat exchanger modelled could be used within any industry where the requirement for heat to be dissipated from electrical components is essential. The parameterisation of the geometry that was completed in the modelling stage allows the dimensions and shape of the heat exchanger to be altered to suit the specifications of any application. This is a major benefit of using command-line based Open Source software.

2. Optimisation of Final Code

The main focus of further research that should be considered is optimisation of the final code that was produced in this report. Initial work has been made towards this goal with a range of variables tested and a clear, concise code has been written to make understanding and editing the code easy for new users. Optimisation of the final code would include: reducing the assumptions made; reducing the number of input conditions required for the analysis; optimisation of the meshing stage and finding an optimum design.

3. Scaling Up of Unit Cell

Analysis of a full-scale model could be implemented in research to ensure the model behaves appropriately at full-scale. This could be done by mirroring and multiplying the unit cell in all directions to create a full model to the desired size required for application. It would be important to check the behaviour of the flow through a full-scale concept compared to the unit cell and confirm there are no unexpected, rogue results.

4. Investigating Other Variables

Other variables such as material selection, input parameters and geometry changes are all areas that could be investigated further in future work. Optimising the design by finding the most efficient selections would prove vital in making the exchanger more practical and more reliable for industry use.

CONCLUSIONS

OpenFOAM is a complex CFD tool, providing very large flexibility due to the ability to modify its source code. It excels in the areas of parameterisation, optimization and automation, however, is not a suitable tool in all cases. Becoming proficient in using it is exceedingly time consuming and requires both a strong fundamental knowledge of CFD as well as excellent fluency using C++ in order to utilise its full potential. It has a logical but complex directory system, and the lack of any GUI and default mesh or solver settings makes it very unfriendly to the user. It also lacks robustness. These features make it a non-optimal tool for simple or small scale simulations, if commercial alternatives are available.