Recently we ran a Summer Lotto and Nick Sibly - one of my ingenious colleagues - decided it would be great to have a proper air-blown Lotto machine - so he built one in his garage! The materials were plywood, a Perspex front panel a couple of steel tubes and an airbed pump from Argos. The balls were polystyrene.
Here is a picture of the machine running ...
The concept was to blow air from the pump located in the base through an inclined tube and see if any balls 'popped up' through the vertical tube at the top of the cabinet. Ideally there would be an occasional ball pop out but not a continuous stream, thus keeping the watching audience (from all our offices around the country on a Skype link) amused and in a state of feverish expectation.
To be fair it worked surprisingly well after a few hours of 'tinkering time'. However, there was one flaw in the design that could not be easily fixed by adjusting flow rate or the tube angles. This was the tendency for the balls to gather on the base in the bottom left hand corner where there appeared to be a 'dead zone'. If your ball was unlucky enough to land there, it was very unlikely that you would ever win the Lotto as your ball would be trapped. You can see this problem in the image above.
Here's where SOLIDWORKS Flow Simulation came in. Fundamentally this is a fluid dynamics problem so I decided to knock up a virtual prototype in SOLIDWORKS and run a Flow test.
Here is my virtual prototype cut with a 'Zonal' cut ...
For those who have not had the pleasure of working with Flow, it is a surprisingly simple and intuitive tool to use. It is fully integrated with SOLIDWORKS and the set up is easy to do. In this case I applied an air flow rate to the bottom tube where the pump was located (using the max flow rate of the pump) and an atmospheric pressure to the outlet. Flow does the rest.
The first project I ran revealed why the balls were trapped. The angled air flow input was creating a region of low velocity in the bottom left hand corner. You can see this in this 'Cut Plot' where the colours represent air velocity ...
The good news was that the plot also revealed that the design would work as there is a significant airflow up the 'chimney' at the top.
This is made even clearer when you see the 'Flow Trajectories' in this animation.
What could we do to improve the design? The answer was to examine the flow trajectories and create a new part that would follow the lines of the airflow. This would blank out the corners and encourage recirculation of the air and avoid the dead zones.
Using the cut plot I created sketches in 2 directions and then use a surface loft to create a smooth shape. I then 'Thickened' and added corners and an inlet tube to match the pump. Here is the design which Nick called "The Pasta Dish" for obvious reasons.
Adding this part to the Flow project revealed an immediate improvement in the results. The air circulated much better and the dead zones disappeared. Here are the improved results ...
Here is the animation of the flow trajectories....
Using our 3D printer we created a physical prototype so we could use it in the real machine. Here is the machine and the part ...
Did it work in practice?
Well take a look at the Videos below and judge for yourself!
What did this mean in practice? Besides having a fairer Lotto machine we also found that the batteries lasted longer as the balls were more likely to exit the chimney, so we saved time and money!
By Andy Fulcher
Solid Solutions Management Ltd