[3dem] LN2 supply?

Bridget Carragher bcarr at nysbc.org
Mon Jan 21 08:38:17 PST 2019 

Thanks to everyone for their generous sharing of information on this topic.  As I promised I am attaching a summary of all replies for the record.
Best regards,
Bridget

Replies received (slightly edited for convenience of reading)
eBIC has been using the direct LN2 feed to 4 Krios in the hall for a while now, and we are quite happy with our system. Will send you specifics directly.
Peijun Zhang
I think the people that do a flavor of Nuclear Magnetic Resonance called Dynamic Nuclear Polarization deal with this issue all the time.
I'm was not involved on the installation of the system at the National Lab I work at, but I can get you the contact of the people involved on it.
Joana Paulino

We pipe the LN2 directly from a big outdoors storage tank into our Talos & Krios. It’s been working fine for 3 years now. The pipes are vacuum-jacketed, so they only need to be cooled down when you first connect the system.The LN2 flows “downhill” from the storage tank, so that gas forming in the pipes will move back up to the storage tank. There is some upper limit to how long the pipes can be without incurring too much loss through gas formation, but I don’t know what it is. So far, our LN2 consumption is not dramatically different from what we would expect from using Dewars.

Safety: we of course have oxygen sensors connected to the ventilation system, such that air (or cool N2 gas) will be evacuated from floor level in case the sensors trigger it. There are a couple of emergency breakers in our control room that will shut off the LN2 supply if pressed.

The scopes will go down if the supply is interrupted, but this has not happened in the 3 years we've been running. Our gas company seems quite used to making much larger installations than ours in, e.g., food processing industries and the like.
Gunnar von Heijne

I am cc'ing Alan Vanderstelt, MSU's project manager for our new cryo-EM facility. We did initially try and plan for direct piping, but in the end we abandoned the idea. We decided to go with 2 nitrogen tanks that can are built together as a single unit so hopefully we will have to change them 2X less often than single tanks.  Alan spent considerable time talking with various vendors to find suitable materials and weigh costs and design issues. He can fill you in on any specific details you are looking for.

I'd be interested to hear the results of this discussion. It's obviously a big issue, so I'm glad you started the conversation.
Kristin Parent


We had such a system when I first came to Wadsworth in 1976.  There was a 2000-liter tank on the loading dock, and diffusion-pumped lines throughout the building.  A company called "Vacuum Barrier" made pumping stations, a combination of a direct-drive pump and a air-cooled DP, which were located at intervals in the trunk line.  We still have a never-used pumping station.

The system worked very well, especially for the HVEM which at the time had a cold trap for every vacuum pump (a total of 18 cold traps!).

The 2000-liter tank sprang a leak, and the lab decided to dispense with the system instead of repairing the leak.  When they did that, we had to use LS250 tanks for the HVEM, which we fed into a 500-liter accumulator, which we still have but no longer use.

I made an auto-filling system for our JEM-3200FSC, using vacuum-insulated delivery lines (but not actively pumped), and vacuum-insulated valves  The lines are about 3-4 meters long, and the system works great -- re-filling is fast, and we can collect data on the same grid for four days.  Without vacuum-insulated lines, the system is just not practical.

Back in the day, we had a fire in one of the TEM rooms because the LN2 supply (from an LS-250 tank) failed, so the solenoid valves strayed open and got so hot that the foam insulation caught fire.  That's why I use vacuum-insulated valves now.  We do sometimes run out of LN2, and there is no problem with overheating.  The TEM cools back down quickly when the LN2 is restored.

We have a (German-made!) switch for automatically connecting a new LS250 if one fails (it senses the tank pressure).  But it either resulted in waste of LN2 (by switching too soon), or it failed to switch at all.  So we gave up on it.

However, our old "Vacuum Barrier" system was very reliable -- it never ran out, and Airgas came once a week to refill it.

Sorry about not answering your specific questions,but I hope the above was of some interest.

Mike Marko

We have an automatic refilling system from an outside big tank. We have had it for 3 years now, same as the microscopes, one krios and one arctica. It never failed,  only once we had the intermediate tank (60L) going a bit under pressure , but that was easily fixed by adjusting a valve.
The system is built in such a way:
A big LN2 which also serves other labs, is connected via pipes running under a street, to an “intermediate tank”, which sits in our microscopes rooms. The big tank is placed 30-40meters away from the intermediate tank. I think that there is a max distance that should not be exceeded in between theses tanks, but I do not know how much. Will ask. There is an exhaust valve ("tube”) releasing outside in the street the excess of gas generated in the LN2 pipes from the big to the intermediate tank. I attach some picture.
The pipes that transfer the LN2 from the intermediate tank to the microscopes have an insulating “vacuum layer”. The Arctica is located very close to the intermediate tank so I was never worried about it. The Krios is instead 7 or so meters away, but still that was never a problem, apart from the fact that it takes 10 minutes longer to refill.

Regarding your specific points:
- The LN2 costs very little, I do not think that there is a considerable rise in price compared to the use of 240L rolling tanks. Most of the cost for us is for the “rent” of the big tank
- In terms of safety I think that in principle the system could be even safer because the maximum that can spill at once is 60L and it takes some time to refill this intermediate tank.Obviously O2 sensors and big extractors are located just by the intermediate tank
- Yes concerns that if the system goes down all microscopes go with it is true. May be it is reasonable  to couple 2-3 microscopes only with one intermediate tank. It is unlikely that something goes bad with the BIG tank. Our is located in the hospital area and I suspect and hope that this is somehow a guarantee that the main gas tanks are well maintained and refilled…

Ah, a convenient thing is that we have a tap for the LN2 always at hand for sample dewars and cells. Let me know if you (or any other) need more info.

Marta Carroni

We tried piping LN2 from our central tank in the adjacent building into our current building (architect vetoed installation of a new tank because it was too ugly. And we had to fight to retain the existing tank for the same reason). We did some informal testing this past summer trying to assess whether we should continue to cart our 240L dewars over to the central tank for filling or use a closer auto-fill station in the new building.

Our experience was that if infrequently used (as our case was/is), recharging the piping (vacuum jacketed for LN2) to cool it down for LN2 cost us roughly twice as much LN2 (based on central tank gauge) when compared to physically carting the 240L to the central tank for filling. The caveat is that we did this testing over a 2 week period in the middle of summer, so it may not cost quite as much LN2 in the wintertime. We did not repeat the experiment this winter. But based on this measurement, we abandoned the idea of directly piping LN2 into the Krios because the cost of infrastructure building (additional vacuum jacketed piping was required) and the premium of LN2 cost to recharge the piping was too high to offset the convenience benefit. I can also imagine if the pipes are not kept cold all the time, that the delay in receiving LN2 might need to be factored into the Krios autofill (maybe have it start fills at 15 or 20%).

We cannot speak to the safety concerns because the central tank fill is in a loading dock and the auto-fill station is in a large lab. Our EHS did not express any concerns regarding asphyxiation via nitrogen-displacement. I believe there was sufficient circulation of air (I do not recall the exact number of cycles per hour required) and was not under positive pressure. Back during my postdoc in Switzerland at EPFL, they were VERY concerned about this. But the Vitrobot was housed in a small lab space. They made us wear oxygen sensor necklaces and there was the ability to hit a panic button in the case of an emergency.

Since we don't directly pipe from the central tank, we operate similar to NYSBC and maintain x2 240L dewars for our 1 Krios. Swapping one out for the other when it gets low. If there is concern that the central tank might fail, then back-up dewars constantly filled with LN2 would likely need to be maintained.

Hope the sharing of our experiences helps.
Paul Shao



I wish we had a problem like managing "6" Krios.
It might be worth talking to Michael Sattler.  I have never seen so many NMR machines in one place before.  I don't know if they have LN2 piping.  Michael Sattler (sattler at helmholtz-muenchen.de<mailto:sattler at helmholtz-muenchen.de>).

Kiyoshi Nagai



Definitely:
Marta Carroni in Stockholm.
Riken Institute in Yokohama.

Maybe:
Christos Savva in Leicester.

Costs 60K.
Wim Hagen



You can contact Daniela Nicastro at UTSW. They built a big tank outside the facility and pipe the LN2 to the microscopes and other usage.
For the safety concern, O2 sensor can solve it (if there is large leakage, the O2 level can drop to <19%, which is dangerous already).
For the cost, I don't know since I wasn't a facility manager when I was there.

One more choice is to directly Airgas or Praxair.
As a professional gas supplier, they should be able to answer your question(s) and concern(s).
We at UT Austin don't use this option, because I wasn't here during the developing stage.

Aguang at UT Austin



I saw your email about piping LN2 to Krios room.   I think University of Michigan have a dedicated LN2 storage piped into their krios room. You may want to contact them to see how well their system works.

Gongpu Zhao


If you haven't heard back from Dan Clare or someone else at Diamond
within a few days, happy to ping him on your behalf - they pipe their
LN2 in from a central location. I assume they watch 3dem, but you never
know.
Alexis Rohou


- we serve: lN2 for 2 microscopes and one lN2-sample storage tank, lN2 faucet for filling dewars, and N2 gas (8 rooms and several TEM/SEMs/DualBeam)
- advantage: very convenient (telemetry and refills of tanks are all automatic), no downtimes due to changing or iced-up 240L tanks, no space wasted for storing tanks.
- the distances for piping are not too large (max maybe 20 m to the Talos)
- costs of installation was about $100k (tank platform with fence and electric (for telemetry), vacuum jacketed pipes, O2 sensors, emergency shutoff valve right outside the building, vents)
- cost of running is about a wash with the costs for 240L tanks: bulk-tank 1500Gal plus small micro-tank for N2 gas supply of the EM facility is rent $740/months, plus lN2 deliveries every 1-2 weeks; we serve more than just lN2 for two scopes (e.g. the cryo-Scios uses quite a bit of N2 gas) => our annual lN2 costs including the tank rent are ~$30k
- for safety: O2 sensors (including crawl-space) with audio and visual alarm and instructions to evacuate area, floor drains also in the crawl-space, raised floors in the scope rooms (provides some time buffer to get to the emergency shutoff valve, 2 emergency shutoff valves (outside the building and at the tank), users receive specific training.
- we are operating it for 2.5 years and have had no issues (knocking on wood)
- one issue that frequently seems to occur is to find a good spot for the tank (distance to facility, distance from building, accessibility for large lN2 trucks ...)

Dany Nicastro


We also have a LN2 pipeline with a self-filling buffer tank in the OIST cryo-EM facility. The system has a liquid-gas separator in front of an internal buffer tank to vent gaseous N2 when the lines are warm. The gas is vented outside. It takes about 10min to cool down the 30 meters of vacuum-insulated stainless steel line from the >5000L external tank (which is filled every 3-4 weeks). The warm-up corresponds to loss of about 50L of LN2 each time the buffer tank is filled from a warm state. The buffer tank is 150L and a lower and upper threshold can be set with an electronic regulator for automatic filling once it gets near empty or full. When the lines are cold, the buffer tank is filled in 15min from near empty. It can also be used as a filling station for pressurized dewars.

In principle, multiple cryo scopes could be connected directly from this buffer tank - this is what it has been designed for. However, unfortunately someone decided years ago that such unattended filling may be too dangerous despite a closed circuit and multiple oxygen sensors with central alarm. So our cryo-TEMs have never been connected and regular dewars are filled instead from the buffer tank at this dispensing point, then shuttled over 10m to the scopes. So although we don't depend on LN2 dewar delivery from an external supplier, we disadvantages: still need to swap dewars, loose 50L each day for filling multiple dewars (not just for cryo-EM), and have danger of exposure to N2 gas when people use it for filling non-pressurized open dewars.
But if the system were directly connected to the scopes as designed, it would be very nice indeed.  I can send specifics if you like. Its price is not justified, however. Should minimize distance to external LN2 tank, and distance from buffer tank to scopes. With 6 scopes, a 300L buffer tank would maximize refilling intervals. Vibration and noise of our system during buffer tank filling are low, but worse when filling dewars from it - it should be installed outside the actual scope room. It is very reliable, no concerns. But must consider local regulations about unattended filling. Bulk delivery of LN2 by tanker is significantly cheaper than trucking in scope-ready dewars. This may compensate for loss due to warming up lines. Bottom line is you want a self-filling buffer tank in the facility, not connect the scopes directly to the same LN2 line. It also minimizes vibration transfer to other scopes when one is on a filling cycle.

Matthias Wolf


We’re happy with the autofill system we have here at eBIC and I recommend it rather than separate pressurised Dewars as a matter of routine use. My biggest concern was the human error factor/H&S issues that swapping so many Dewars in one room introduces and the autofill system mitigates a lot of these. We can swap back to Dewars on short notice as needed and this is also important in the event of a failure.
I’ll put some details together and get back to you by mid week.

Alistair Siebert



Cryo-EM facility at SUSTech has been using pipe to fill in its 6 Krios and other instrument with total 9 filling points. A design term for our system only has experiences for industry and sysnchrotron.  Our experiences so far are as:
1, Cost for LN2 is high than use tank directly. This is because every time filling in LN2 to any scope's buffer tank, all pipe have to be cold down. Even installed an isolate valve for all filing points, it still needs to cool down all pipe when a far filling point needs to be used.
2, LN2 vaporisation is a big issue when filling in tank take place. For such large amount of N2 gas it raises a serious safety issue in the room.
3, The rate of LN2 evaporation of pipe has to be considered when these pipe have been cold down for all time, especially using a fairly length pipe.
4, It is very time consuming to cool down if pipe do not keep in LN2 temperature all time.
5, The noise is very loud during the pipe cooling down cos N2 gas has to be released.
We are now modifying our system. Hopefully, these issues will be resolved.
Peiyi Wang



I can tell you that we considered, and rejected, the idea of filling our Xray generator's cryo-dewar from the NYSBC bulk LN2 tank.  That dewar has (had?  I'm not sure if it's still there) a 40 L capacity, and would have been topped up 1-2x per day during continuous use of the CryoJet it was attached to.  I'm not sure how that compares to the dewar of a Krios, but I suspect it's not far off.

Our main concern was not safety or disturbance to the area (although we probably should have thought about that more), but rather the huge boil-off needed to cool down the vacuum-jacketed pipes before liquid would come out the far end.  Not only does this waste LN2, but we would have needed some clever piping to avoid having the nitrogen gas spraying directly into the CryoJet's dewar.  Since that gas would initially be close to room temperature, this would cause a lot of LN2 loss in the dewar as well, potentially compromising the flow of LN2 to the CryoJet.

Matthias Wolf's suggestion of an intermediate dewar might be the solution you need, but it seems to add complexity and chances for failure.  I might change my tune if I had three Krios to feed, however!

Matt Franklin




End

On Jan 18, 2019, at 9:27 AM, Bridget Carragher <bcarr at nysbc.org<mailto:bcarr at nysbc.org>> wrote:

Hello all,
We are building a new microscope suite for 3 Krios to add to the 3 we already have.  As everyone who does this will know, managing the LN2 for these instruments becomes quite a burden in terms of swapping tanks, and managing storage and exchange of these tanks.  So we are considering piping LN2 directly into the rooms.  Can anyone with experience please comment on:
- Costs of LN2 when considering that the pipes needs to be cooled down prior to fill and the pipe runs are fairly lengthy from the central storage tank to the rooms.
- Safety concerns in terms of filling accidents in the room displacing oxygen; is this a major problem compared to the 240L tanks
- Concerns that if anything goes wrong with the central storage unit all 6 instruments would go down
- Anything else?
If people want to send their comments directly to me I will collate and share with the list after a few days.
Thanks,
Bridget

---------------------------------------------------------------
Bridget Carragher
Simons Electron Microscopy Center
New York Structural Biology Center
89 Convent Avenue, New York NY 10027
(212) 939-0660;  bcarr at nysbc.org<mailto:bcarr at nysbc.org>
National Resource for Automated Molecular Microscopy
http://nramm.nysbc.org
----------------------------------------------------------------



---------------------------------------------------------------
Bridget Carragher
Simons Electron Microscopy Center
New York Structural Biology Center
89 Convent Avenue, New York NY 10027
(212) 939-0660;  bcarr at nysbc.org<mailto:bcarr at nysbc.org>
National Resource for Automated Molecular Microscopy
http://nramm.nysbc.org
----------------------------------------------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ncmir.ucsd.edu/pipermail/3dem/attachments/20190121/99016932/attachment-0001.html>

More information about the 3dem mailing list