Refine venturi design using computational models

[martukas]

Before going production processes, all designed parts will be analysed and this will cause more accurate product after design. Team will collect datas which will be close to the experimental datas without starting any production especially in Venturi Design.

This work will help us to foresee our product outcome, and we can define working prenciple more specifically such as controlling of respiration. We(Yusuf and Mert) think that we can provide all of this datas to the team. also we have some datas which related with maximum working condition of Venturi Design and the values we get are close to real-time values. All documentation will be below with detailed explanation soon.

This analysis contains multiple inlet variables, this variable reflects behaviour of flow inside the tube. Also it helps us to reveal all necessary data to foresee what may we encounter if we decided to produce part.

We checked if the existing model complies with these conditions by using ANSYS CFD analysis. In addition we saw that existing model was tested with these conditions. So we found a chance to compare our analysis parameters with real-time test. We have seen that our simulation results are really close to real-time results. Existing model had too much turbulence that was affecting the outlet flow. The increase of the velocity in venturi region was creating this turbulence. Also the pressure in venturi region was dropping too much. Then we started to work on new design and new analysis on the data.

[martukas]

Previous work on this question was done in #705

Other related tickets:

631

1122

636

[martukas]

Data files should be dumped on Google drive

[yusufcance]

We did not want to change anything with the outer design because important design requirements were related with inside. To overcome with high speed and low pressure in venturi side, we operated part wider. The reason we did not make exact circle is to send more stable air for sensor to analyze.

[yusufcance]

[yusufcance]

After creating new model, we put it to the parametric analysis to gather different result in different inlet conditions. Result on models are as fallows with less eddies, turbulence, and less difference between inlet and outlet by side of speed and pressure.

[yusufcance]

[yusufcance]

After all of these necessary informations and related datas we got from ANSYS, we would like to prepare plot which consists data of Pressure Difference on Sensors-mass Flow Rate on small region of tube. This analysis shows us we improved design by making a few changes inside of tube. This helps us to get close values im real-life tests.

[inceptionev]

Results from the venturi comparison with the original design. Data is at this link: https://docs.google.com/spreadsheets/d/1CylfLKrE64rkY6gbZZj0AzY_0tv3djxfM-mScOxUXCw/edit?usp=sharing

• The modified design produces less restriction of the flow, so the maximum flow is higher when the blower is turned up to its highest setting (9V for this setup)
• The sensitivity of the modified design is lower than the original, but neither get close to the full range of the 4kpa (40cmH2O) range of the dP sensor used in the ventilator design, and thus resolution is sacrificed, while also imparting a high pressure drop to the supply path of the ventilator
• This indicates to me that we should consider a lower-pressure dP sensor and a lower sensitivity venturi design to both reduce pressure drop and increase resolution.

[inceptionev]

Here is a plot summarizing what performance goals we want for the venturi:

[yusufcance]

Real-time value helped us to analyze our values in a more realistic way which can be easily compared with the computational value. As you can see from the graph upward, the value we got from the new design has approach the expected value when the inlet flow increases. But main target is to approach the curve as much as we can. It requires different modeling techniques to understand the outcome. Understanging of the flow here is the key factor because to analyze flow you have to understand the flow path, affects which are created by flow also the understanding of the venturi principle. To claim expected curve, we have used 5 different models and put them the ANSYS Fluent to see how flow get affected with different modeling techniques.

[yusufcance]

1) Slot Venturi Zone with Regular Outlet Pipe

[yusufcance]

2) Round-Small Venturi, Half-Regular Inlet and Regular Outlet

[yusufcance]

3) Smaller Venturi, Fully Regular Pipe

[yusufcance]

4) Smooth Transiton

[yusufcance]

5) Modified Smooth Transition

[yusufcance]

Those are 5 different inner designs and and flow analysis result. We gather all of the data to create parametric results. We had 2 real-time data, base and first modified design, which are not sufficient to create fully accurate parametrization. But with 5 new different data which are created in order will be very helpful and useful to create other options. Our main expectation is to get results from fifth design which may be more effective and accurate.

Theory of this design is to reduce rupture in the flow. By keeping less slope in the transition and we mostly provide that target. On the other side, we tried to keep clear sensor port inlets, that’s why there are difference in the first sensor pipe inlet between fourth anf fifth model. This ideology prevents air squeeze that helps us to get more accurate data. One other advantage is using smooth transition makes mathematical calculations easier. Because if you create more than one radius in the same curve, it will be harder on calculations. Analysing programs such as ANSYS create lesser need on mathematical calculations but to be certain it is also good to check mathematical side of the Venturi Principle.

[inceptionev]

Measured data from 3D printed #5: Modified Smooth Transition:

Data link: https://docs.google.com/spreadsheets/d/1CylfLKrE64rkY6gbZZj0AzY_0tv3djxfM-mScOxUXCw/edit?usp=sharing

[inceptionev]

possibly useful paper on venturis: https://sci-hub.se/10.1016/j.petrol.2011.08.008