Splashing of the water in vacuum cleaner pan caused by the air flow, which goes through the water.

Hyla is a Slovenian company that produces water filtration vacuum cleaners, which are selling in direct sales. They are selling the product all over the world and they are the leading manufacturer in this field. The product is designed, developed, tested and manufactured in Slovenia.

The core of our vacuum cleaner is separator. Separator is a small fast rotating element, which is attached directly on the shaft of electromotor. Its main goal is to separate the water droplets and dust particles from the airflow.

Currently we are developing a new generation of separators. We have problems to figure out how the presence of water droplet on the separator blades helps us separate the dust particles from the air flow. We would like to know the size distribution of the water droplets that are generated by splashing the water which is caused when the airflow goes through the water. In the past we were trying to do such simulation with Ansys but we did not have enough computer power.

We hope that we can get the results from the simulation during this project.

The company Hyla d.o.o. has introduced the problem involving extensive fluid dynamic simulation. The intention was to simulate a three phase (fluid, air, dust particles) flow in the vacuum cleaner and furthermore improve the design of the device. The discussion has followed regarding the magnitude of such problem in physical and numerical sense. It has been concluded:

The preliminary simplified numerical case that was run on desktop computer should advance to more complex schemes and be promoted to HPC.

A vast computer resources are expected to be required, since three phases, turbulence and fairly complex geometry are involved  in this case.

The numerical case that combines all necessary physical attributes should be prepared. It is expected that the setup of such case should take 3-4 months to prepare.

All involved parties have been in contact ever since and further discussions regarding the numerical problem setup are in progress.

The conclusion was that, the setup of the numerical case is fully prepared, the assessment will be made, firstly about the scaling of the case on multiple HPC cores at ULFME and then the first optimization iteration of the case will follow to lower the computational cost of the numerical case.

Currently there is only an assumption of the amount of computational cost for this case to be solved which will much exceed the computational cost that is currently available.