MAGNETS AND PUMPS DON'T MIX: AN ADVENTURE TALE
(received 12/10/90)
6 December 1990
Dear Professor Shapiro,
An Oxford 1.9T/310mm horizontal superconducting magnet (#B26694) has been a workhorse for the past six years in our laboratory. On October 26th, the evening before the 18th (NMR)", New Mexico Regional NMR, meeting and shortly after receiving a "Reminder" from you, it vigorously ingested a 17-pound Cole-Farmer Masternex peristaltic pump (Model # 7523-00). After glancing off the heavy end plate, the pump caught the shim/gradient coil and displaced it about 60 cm before coming to rest in the center of the magnet.
After the initial shock wore off, and after a sufficient number of people gathered to survey the situation, an exploratory effort was made to extract the pump by applying an axial force to the pump through a rope (yellow, polypropylene, 10 mm diameter). This effort was terminated after we moved the entire magnet 10 mm because we realized that there was a real danger of 1)pulling the magnet off the wooden pallet and 2) of displacing the solenoid with respect to the dewar.
At about this time, a serious thought was given to discharging the magnet in order to remove the offending pump but it was rejected on the grounds that 1) it was the sensible thing to do and 2) we never acquired the magnet power supply as we had hoped so there would be a significant delay before recharging the magnet. Therefore, we opted to take some risks, but not enough to cause any damage to the solenoid. At this stage, there were no overt damages to the magnet; coolant boiloff rates were normal and the magnet seemed to be working but probably with a less than optimal shim.
Another suggestion was to somehow separate the major components of the pump so they could be removed piece by piece. After some strenuous efforts at wielding a magnetic hacksaw blade and a magnetic screwdriver in the bore, we abandoned this effort.
Reasoning that a lot of harm could result from exerting forces between the solenoid and the dewar, we decided to try a scheme to let the solenoid push out the pump by pulling in lots of smaller ferromagnetic items that could be removed from the magnet individually after the pump was removed. After protecting the fiberglass bore tube with a 25.4 cm diameter PVC water pipe that enclosed the pump, we built a 16 cm long wooden piston that fit tightly in the PVC pipe and located it adjacent to the pump. We then confined a bundle of 16 120 cm long and 19 mm dia. PVC water pipes to the center of the large PVC pipe and placed it next to the piston. Finally, in a scene similar to that of the operation of some early nuclear reactors, we inserted 5-foot long, 12 mm diameter steel "rebars" one-by-one into the 19 mm PVC pipes, starting at the center of the bore.
Initially, the process seemed to be working. However, after the pump had been displaced about 30 cm, it became impossible to keep the pump centered in the bore and we finally appreciated that the radial force on the pump was formidable when the pump was displaced from the uniform field region. As we were contemplating our next move, we noticed an alarming increase in the liquid helium boiloff rate. The steel rebars were extracted immediately to reposition the pump in the center of the magnet but it was decided to quench the magnet because we could not afford the wait to see if the boiloff would calm down before we ran low on liquid helium without any supply on hand. During the gentle quench, we used up less than 20 liters of liquid helium to end up with about 17 liters in the helium can. The pump was removed without much fanfare. We blew out the coolants, warmed the magnet, and disassembled the dewar to discover that the various cans were reasonably centered so that just removing the pump and possibly also warming the dewar was probably sufficient to correct the enhanced boiloff. However, we were glad to make certain that the cans were centered and also discovered that the fiberglass rods that kept the helium can and the radiation shield centered in the nitrogen can were looking more like brooms than rods. We replaced those spacers and recentered all cans on reassembly. In addition, we inserted 15 cm by 100 cm multilayer pads of aluminized mylar superinsulation and applied them under the joints between the superinsulation in the vacuum space around the girth of the nitrogen can and the superinsulation pads applied on the annular end plates bolted to the nitrogen can.
We pumped out the vacuum space, precooled the magnet with liquid nitrogen, blew it out into the nitrogen can after two days, waited overnight to boiloff residual nitrogen (with appropriate pumping of the helium space), leak checked one last time, and finally cooled with liquid helium before powering back up. The only other fix we had to perform was to work on the output transistors for z2 and z3 shims which blew when the shim/gradient coil was displaced by the pump.
Besides reporting that we have extricated ourselves from a brink of disaster, we wish to pass on some other good news. First is that for the second time in six years, we allowed a cleaning crew to clean and wax the floor while the magnet was down. The second is the fact that our helium consumption is back to where it was (we buy a 100 liter dewar of it every 6.5-7 weeks) and the nitrogen consumption has been reduced from 8% to less than 6% per day, presumably because of the extra superinsulation.
Somehow this adventure reminds us of a quote attributed G. L. Mallory "…because it is there."
We acknowledge the help of Jim Carolan of Nalorac, Dick Marsh of Oxford, and Neal Munro of Magnex. We also thank Clif Unkefer and Scott Ekberg of Los Alamos National Laboratory for the loan of an Alcatel leak detector and James Abbott of MIT and Los Alamos for overseeing its transportation. Finally, we thank Leon Axel for not only lecturing at the (NMR)2 meeting but also for suggesting the title for the would-be picture (not available for this letter) of the magnet containing the pump.
Sincerely yours,
Steve Altobelli, Arvind Caprihan, Eiichi Fukushima, Milt Icenogle, and Paul Majors