wrapping it up
Two days ago we spoke with Dr. Figueroa once more. He recommended that we break up the lower mass of our L-shaped conventional shield into several thinner washers instead of one large one, and also that we put rings around the outside of the conventional shield at the top and the bottom of the L. He said to experiment with both Cryoperm and Vanadium Permendur, and then to simply write a report of our progress on Friday. We have done so, and found that both recommendations improved the overall attenuation/mass efficiency of the shield by a great deal.
a little bit of direction
Despite the fact that we solved the singular matrix problem, we have still been having problems solving our model. Our faculty advisor Dr. Brad Minch poitned out that many non-linear solvers need functions that have smooth derivates otherwise they become unstable. Sure enough, when we smoothed out our functions for permeability the solver began to converge. Finally we are able to get some results and we should try to contact Dr. Figueroa early next week.
Finally we are able to figure out how to get the saturation working in COMSOL. We were having difficulties with a singular matrix error when we try to solve the problem with a permeability function. Apparently, when implementing a non-linear function for a material property, we need to use the non-linear solver. Who would have thought?
Yesterday we met with our NASA advisor, Dr. Enectali Figueroa, and showed him our current shield model. It seemed to be working well, and the detector plate was more than sufficiently shielded. He experimented with the model and told us that it was not modeling saturation at all. He told us to implement saturation with a permeability function, edited in such a way that once the material should saturate in reality, the function sets its permeability to 1, so that the shield acts like air.
He thought it would be possible to model the saturation of a superconducting shield, but it would probably be much more difficult to implement in COMSOL, so we shouldn’t bother.
lots of progress
No, we're not dead, just too busy working to update the site! Over the last week we put together an 5 degree of freedom optimization script to determine the geometry of the conventional magnetic shielding with the highest attenuation and the best attenuation to mass ratio. When this was completed, we combined these results with the superconducting shield work and got a first pass at the full shielding design. We are planning on talking with Dr. Enectali Figueroa early next week to discuss this model and where it can be improved.
began superconducting shield
With a preliminary magnetic shielding model completed, the group is now able to split its efforts and tackle the superconducting shield as well as the conventional magnetic shield. The group will use the most recent results from the high permeability shield and create a superconducting shield that will reduce the field to the necessary 10-7 teslas.
began conventional magnetic shield
With the new tutorials in hand, we were then able to implement a saturation regime in FEMLAB. We created a simple enclosure around the magnet and began exploring the effects of saturation and geometry on the magnetic field strength.
We had difficulty implementing a magnetic saturation point in FEMLAB/COMSOL even with assistance from Olin Professors Rebecca Christianson and Steve Holt. Our next course of action was to call the company and ask them for the best way to implement this. We were directed to a series of tutorials in a different version of the software and they proved to be quite helpful.
After his conference at Princeton, we were finally able to visit Dr. Enectali Figueroa in his lab at MIT. He gave us some better insight as to the nature of our project and helped us get started.