## Heat exchanger

Threads relating to the design of liquid-fluoride reactors.
Lars
Posts: 3060
Joined: Jul 28, 2008 10:44 pm

### Re: Heat exchanger

mlippy38 wrote:Wow, \$35M. That is even larger than I was expecting, and is particularly encouraging for my design.
Yes it is a lot. But keep the other costs in mind. If you can save 1 m^3 of fuel salt that is worth \$3 to \$7M depending on the fissile concentration at startup - and has other fairly important side benefits too. Likewise, assuming we are using a Brayton cycle then reducing the Thot_in - Thot_out by 50C would be worth \$17M/year or roughly \$140M!!! As David mentions, for the first units we may well go with steam just to save on up front R&D expenses in which case we can't use the higher Thout_out anyway but in the future it will be worth a whole bunch of Hastalloy-N to reduce the Thot_in-Thot_out.
- If you lengthen the heat exchanger infinitely, the two fluids will reach an equilibrium temperature (approach temperature equal to 0). Thot,in - Thot,out would be very large.
Won't it be nearly zero in this case?
If you make the HX very short, it won't give the streams room to exchange heat. Thot,in - Thot,out would be minimal.
In the extreme, won't Thot,out = Tcold,in if the HX was essentially zero length?
This comes from the "A" term in q = U*A*Cmin, where A is the heat transfer area (perimeter*length).
- U comes from the convection terms of each fluid and the material by which they are separated. 1/U = [(1/hc)+(1/hh)+(t/k)] where hc is the convection HT coefficient of the cold fluid, hh is the convection HT coefficient of the hot fluid, t is the thickness of the boundary separating the flows, and k is the thermal conductivity of the boundary material. The convection terms are an interesting study themselves, as they reach peak values in the transition from laminar to turbulent flow. Due to the high pressure drops of the small channels already existing in compact heat exchangers, laminar flow is maintained to limit the issue from worsening. Turbulent flow always provides better heat transfer (higher hc and hh), but the price of dramatically higher deltaP is prohibitively high.
It will be interesting to see where you land. Do you have the hc,hh, and t/k numbers for the MSBR design? I recall reading where they knurled the insides of the tubes to increase the turbulence.

I'll tackle understanding Nusselt numbers another day - I just came off of a very long work shift.
Thanks,
Lars

mlippy38
Posts: 58
Joined: Feb 12, 2010 5:07 pm
Location: Blacksburg, VA

### Re: Heat exchanger

Lars wrote: Won't it be nearly zero in this case?
No! The longer the HX, the more heat will be continually removed. Heat will be removed from the hot fluid is at the same temperature of the cold fluid coming in. At that point, Thot,in – Thot,out is very large.
Lars wrote: In the extreme, won't Thot,out = Tcold,in if the HX was essentially zero length?
No, Thot,out will equal Thot,out. No heat will be exchanged, and all boundary conditions will hold true.
Lars wrote: It will be interesting to see where you land. Do you have the hc,hh, and t/k numbers for the MSBR design? I recall reading where they knurled the insides of the tubes to increase the turbulence.
I attached a Word document with some screenshots from ORML-TM-1545 that have information on the heat transfer coefficients and other characteristics of the MSBR Primary HX.
Attachments
Screenshot from ORNL-TM-1545.docx

mlippy38
Posts: 58
Joined: Feb 12, 2010 5:07 pm
Location: Blacksburg, VA

### Re: Heat exchanger

David wrote:Matt, Lars,
Ignatiev's 2007 paper has a nice break down of tonnes of Hastelloy and costs (not sure how dated the numbers are though).
David L.
Thanks for posting that, David.

Here are some pieces of information I found interesting:

-Tubes of Hastelloy N cost 3x as much as sheets. Fabrication cost for shell-and-tubes seems awfully high.
-“From them, in particular, followed, that cost of primary and secondary salt carriers was insignificant compared to the total capital investments in the plant, whereas expenses for components made from Hastelloy NM (29 %) and graphite (6 %) made about 35% from the total capital investments.”
-“The analysis allows planning a actions directed on reduction of cost for MOSART primary and secondary systems equipment. First of all, it is represented expedient instead of shell-tube heat exchanger to use plate type heat exchanger, made of foffered sheets with 1mm thickness. In such heat exchangers higher specific heat removal and cheaper sheet material allow to decrease cost considerably.” – It looks like they are thinking along the lines I am.
-“…components made of Hastelloy NM cost of an alloy itself made about 33 %, and 67 % of cost concern to expenses for manufacturing of the main components of primary system.” – Very important, and not what I was expecting. For as attractive as my design seems from a volume and material reduction standpoint, the simplification of manufacturing may actually be the key aspect.

Lars
Posts: 3060
Joined: Jul 28, 2008 10:44 pm

### Re: Heat exchanger

mlippy38 wrote:
Lars wrote: Won't it be nearly zero in this case?
No! The longer the HX, the more heat will be continually removed. Heat will be removed from the hot fluid is at the same temperature of the cold fluid coming in. At that point, Thot,in – Thot,out is very large.
I see I misunderstood the meaning on Thot,out (I thought of it as the coolant exiting the HX). I'll go back and reread the previous posts with this in mind.

So thing I'm most interested in is maximizing the coolant output temperature (Tcold,out) while keeping the fuel salt volume minimum and not getting out of hand with the pressure with the constraint that the fuel salt maximum temperature (Thot,in) is fixed. The fuel salt return temperature (Thot,out) has a floor around 560C and likely that is where we want to operate to reduce the flow required and hence the pumping power and pressure.

Thanks,

Ida-Russkie
Posts: 490
Joined: Aug 29, 2008 4:55 pm
Location: Idaho Falls, Idaho

### Re: Heat exchanger

-Tubes of Hastelloy N cost 3x as much as sheets. Fabrication cost for shell-and-tubes seems awfully high....

-“…components made of Hastelloy NM cost of an alloy itself made about 33 %, and 67 % of cost concern to expenses for manufacturing of the main components of primary system.” – Very important, and not what I was expecting. For as attractive as my design seems from a volume and material reduction standpoint, the simplification of manufacturing may actually be the key aspect.
I was talking with an H&V Engineer that used to work at the Idaho chemical processing plant. He was telling me some of the corrosion problems they have had over the years in their evaporator. The process used some acids which tended to concentrate chemicals from the process. They also added chlorine to fight algae in the fuel storage pools. What they would do is buy the whole heat from the foundry of Hastelloy. When their venders needed material the foundry was authorized to make them bars, plates or whatever from this stock. He said when the evaporator failed it developed pin whole leaks in the welds.This is because they had not developed a proper welding rod for the Hastelloy they were using.

So yes a simple to build design with as little welds as possible.

Cyril R
Posts: 5037
Joined: Jul 14, 2008 3:12 pm

### Re: Heat exchanger

Ida-Russkie wrote:
-Tubes of Hastelloy N cost 3x as much as sheets. Fabrication cost for shell-and-tubes seems awfully high....

-“…components made of Hastelloy NM cost of an alloy itself made about 33 %, and 67 % of cost concern to expenses for manufacturing of the main components of primary system.” – Very important, and not what I was expecting. For as attractive as my design seems from a volume and material reduction standpoint, the simplification of manufacturing may actually be the key aspect.
I was talking with an H&V Engineer that used to work at the Idaho chemical processing plant. He was telling me some of the corrosion problems they have had over the years in their evaporator. The process used some acids which tended to concentrate chemicals from the process. They also added chlorine to fight algae in the fuel storage pools. What they would do is buy the whole heat from the foundry of Hastelloy. When their venders needed material the foundry was authorized to make them bars, plates or whatever from this stock. He said when the evaporator failed it developed pin whole leaks in the welds.This is because they had not developed a proper welding rod for the Hastelloy they were using.

So yes a simple to build design with as little welds as possible.
For aquous chemistries involving chlorination and chlorides all at the same time, few choices are available. This is the absolute worst in terms of corrosion. Titanium alloys do rather well though. They are fancy and expensive but are one of the few choices available. Pulse chlorination is a good way to kill stuff and it is widely considered best available technology (BAT) so will be licensed most of the time. We ordered some titanium tubing for condensor test setups. We want to know about flow assisted corrosion using high and frequent chlorine loadings. This was months ago - it still hasn't arrived and the supplier says wait 'up to two years'!! Be prepared to wait a long time. This stuff is in high demand and there are relatively few qualified manufacturers.

Needless to say the LFTR will have to keep oxygen species out. And use redox control to keep the free fluorine busy at dinner. There is still the condenser if using seawater cooling. New UV based biofouling treatments seem very promising. No nasty chemicals. Well fewer at least.

Lars
Posts: 3060
Joined: Jul 28, 2008 10:44 pm

### Re: Heat exchanger

Suppose we have Thot_in at 704C, Thot_out at 593C, fuel salt at 4.4e6 J/m^3-K, and a 2.4e9 Watt (thermal) reactor. To move the heat the salt flow must be 4.9 M^3/sec. Suppose that the core is 3m in diameter, has a 1 m thick blanket, and a 0.3m thick neutron absorber. Further imagine that the ideal flow rate inside of the HX is 2m/s (I really don't know what it is for this HX). We then need 2.46m^2 cross-sectional area for the fuel salt in the HX. If the HX is 3m high then the fuel salt volume is 7.4 m^3. If 30% of the total HX cross-section is fuel salt area and the HX hugs the reactor then the HX is around 0.5m thick.

So a key question is what is the ideal flow rate for the HX with 2mm spacings between plates? And what would the pressure drop be for 2-3 meters of such a HX? We may be lucky and get an acceptable answer with simple top to bottom fuel flow (and bottom to top for the coolant).

mlippy38
Posts: 58
Joined: Feb 12, 2010 5:07 pm
Location: Blacksburg, VA

### Re: Heat exchanger

Lars wrote:Suppose we have Thot_in at 704C, Thot_out at 593C, fuel salt at 4.4e6 J/m^3-K, and a 2.4e9 Watt (thermal) reactor. To move the heat the salt flow must be 4.9 M^3/sec. Suppose that the core is 3m in diameter, has a 1 m thick blanket, and a 0.3m thick neutron absorber. Further imagine that the ideal flow rate inside of the HX is 2m/s (I really don't know what it is for this HX). We then need 2.46m^2 cross-sectional area for the fuel salt in the HX. If the HX is 3m high then the fuel salt volume is 7.4 m^3. If 30% of the total HX cross-section is fuel salt area and the HX hugs the reactor then the HX is around 0.5m thick.

So a key question is what is the ideal flow rate for the HX with 2mm spacings between plates? And what would the pressure drop be for 2-3 meters of such a HX? We may be lucky and get an acceptable answer with simple top to bottom fuel flow (and bottom to top for the coolant).
In order to keep a reasonable pressure drop, flow must be in the laminar regime throughout the HX. Transition Reynolds number (where flow begins to become turbulent) is around 2000-2300, so I have used 2000 to stay conservative. Just to be clear, this analysis needs to be done on one single channel. From there, the total flow rate needed can be divided among N number of channels to remove the requisite amount of heat.

Re = (rho*V*Dh)/Mu where rho is the density of the fluid, V is the velocity of the fluid, Dh is the hydraulic diameter of the channel (4*channel area / channel perimeter), and Mu is the viscosity of the fluid.

This gives us a maximum velocity, which can then easily give us a maximum mass flow rate.
As for the pressure drop, I am going to rely on CFD results as well as known correlations. General results for any internal flow can be found using the Fanning friction factor, density, hydraulic diameter, and some constants.

Correlations are available for compact heat exchangers using microchannels and for rectangular channels. Numerous papers have been written solely on the validity of those correlations and sensitivity analyses. I can look through my notes tomorrow and post some of these if you’re interested. This will actually be good for me too, as I’ve been meaning to look back at some of those books that had really interesting information.

I also have an Excel document that I have been using for my simple calculations. I can post it if you’d like to take a look. Everything is labeled nicely, but it still may take some work to fully understand all the formulas and such that are at work behind the scenes. The dimensions I am using for my CFD simulations are based on results from that Excel document, where I tried to keep the pressure drop approximately equal to that of the MSBR shell-and-tube design, and calculated the SA/Vol and other characteristics. It's pretty useful and interesting to be able to change the channel height, channel width, number of channels, channel spacing, and other characteristics to see the sensitivity of each variable.

iain
Posts: 501
Joined: Jan 10, 2007 5:09 pm
Location: Los Altos, California
Contact:

### Re: Heat exchanger

Are you guys talking about the same kind of heat exchanger? Can you both reply if this is what you are referring to?
I suspect Mark may not be looking at a counterflow heat exchanger, and I suspect that Lars is talking about a counterflow heat exchanger.

I've confirmed Lars' calculations of the value of having a low temperature drop from Thot,in to Thot,out. Here's my spreadsheet. The summary, however, is that the primary and secondary heat exchangers have the same net present value associated with temperature drop, and the tertiary heat exchanger, because it is at the lower temperature end of the Carnot cycle, has a much higher value associated with temperature drop. And, all these values are enormous (the units on the vertical axis are dollars -- that's hundreds of millions of dollars). You really want to spend big bucks on the heat exchangers.
image (2).png (8.43 KiB) Viewed 2126 times
Note that the value of the heat exchanger is linear with temperature drop, whereas the cost of achieving that temperature drop will be hyperbolic. There will be a well-defined optimum point where those curves have the same slope.

Jess Gehin made a comment to me earlier this week: primary pumps (as opposed to primary convective flow) are a really valuable feature, because they allow the primary HX to have lower delta-T.

Also, note that higher efficiency primary and secondary HXs have an interesting side effect not included in this trivial analysis: as heat drop goes down, carnot efficiency rises and the rejected heat flow goes down, which reduces the temperature drop across the tertiary heat exchanger (assuming that piece of equipment stays constant), and gives you a little more bang for the buck. Not a huge effect, but I bet it's worth a few million dollars per plant.

-Iain

Lars
Posts: 3060
Joined: Jul 28, 2008 10:44 pm

### Re: Heat exchanger

We both are talking about the a pure counter flow heat exchanger. I just messed up the jargon being a novice in heat exchangers.

Fuel salt in = Thot_in
Fuel salt out = Thot_out (I see I'm not the only one for whom this nomenclature was counter intuitive)

Coolant salt in = Tcold_in
Coolant salt out = Tcold_out

So using the jargon of the field it is Thot_in to Tcold_out that is key.

I really would like to be able to deploy without using the ocean or rivers for cooling. My expectation is that such cooling systems are the achilles heel for fast deployment. At any time prior to starting the plant you are very vulnerable to a political decision to prevent it being turned on using the fig leaf of cooling water impact on the environment. Even once the reactor is running it is still a threat. So, I'm expecting we will have to live with a Tcold that isn't as cold as we'd really like. Hence part of the desire to really watch the temperature drops on the top end.

Lars
Posts: 3060
Joined: Jul 28, 2008 10:44 pm

### Re: Heat exchanger

mlippy38 wrote:
I also have an Excel document that I have been using for my simple calculations. I can post it if you’d like to take a look. Everything is labeled nicely, but it still may take some work to fully understand all the formulas and such that are at work behind the scenes. The dimensions I am using for my CFD simulations are based on results from that Excel document, where I tried to keep the pressure drop approximately equal to that of the MSBR shell-and-tube design, and calculated the SA/Vol and other characteristics. It's pretty useful and interesting to be able to change the channel height, channel width, number of channels, channel spacing, and other characteristics to see the sensitivity of each variable.
Yes I'd be interested in seeing the spreadsheet.
Iain posted things to google docs earlier - I've just started using it and it is a pretty nice way to communicate remotely like this.

iain
Posts: 501
Joined: Jan 10, 2007 5:09 pm
Location: Los Altos, California
Contact:

### Re: Heat exchanger

Matt,

I just read the doc. Great stuff. The pictures help a LOT.
- Have you considered how the fluids get in and out of the channels? I looked at a slot design earlier, and found this to be a source of great complexity and flow restriction.
- I also like the slot design!
- Consider decreasing your channel length by 10x, and increasing your flow cross section by 10x. Same area. Note that the resistance to flow goes way, way down. Car radiators push this very var, and have channel lengths for the air stream which are a small fraction of the sqrt(flow cross section).
- I like your idea of putting the HX around the core and using it to route fluid from the top to the bottom of the core -- this minimizes out-of-core volume. Now consider what happens if you put the HX INSIDE the tank holding the core fluid. Ignore the irradiation problems for a moment. You end up with a tub of molten salt, with radiators around the periphery, slow up flow in the center and faster downflow between the radiators and the outer wall.
- Now consider the radiation problem on the HX. How can you shield the HX?

-Iain

USPWR_SRO
Posts: 489
Joined: Apr 24, 2008 4:54 am
Location: Columbia, SC

### Re: Heat exchanger

Lars wrote:We both are talking about the a pure counter flow heat exchanger. I just messed up the jargon being a novice in heat exchangers.

Fuel salt in = Thot_in
Fuel salt out = Thot_out (I see I'm not the only one for whom this nomenclature was counter intuitive)

Coolant salt in = Tcold_in
Coolant salt out = Tcold_out

So using the jargon of the field it is Thot_in to Tcold_out that is key.
FWIW, in the commercial industry, Thot always refers to the outlet of the Reactor, Tcold refers to the inlet of the Reactor.

Any other HX the hotter fluid is referrred to as the primary side, the colder fluid as the secondary side. So you would have a Primary Inlet, Primary Outlet, Secondary Inlet, Secondary Outlet.

So in this case:
Fuel salt in(to the HX) = Primary Inlet
Fuel salt out (from the HX) = Primary Inlet

Coolant salt in = Secondary Inlet
Coolant salt out = Secondary Outlet

mlippy38
Posts: 58
Joined: Feb 12, 2010 5:07 pm
Location: Blacksburg, VA

### Re: Heat exchanger

USPWR_SRO wrote:
Lars wrote:We both are talking about the a pure counter flow heat exchanger. I just messed up the jargon being a novice in heat exchangers.

Fuel salt in = Thot_in
Fuel salt out = Thot_out (I see I'm not the only one for whom this nomenclature was counter intuitive)

Coolant salt in = Tcold_in
Coolant salt out = Tcold_out

So using the jargon of the field it is Thot_in to Tcold_out that is key.
FWIW, in the commercial industry, Thot always refers to the outlet of the Reactor, Tcold refers to the inlet of the Reactor.

Any other HX the hotter fluid is referrred to as the primary side, the colder fluid as the secondary side. So you would have a Primary Inlet, Primary Outlet, Secondary Inlet, Secondary Outlet.

So in this case:
Fuel salt in(to the HX) = Primary Inlet
Fuel salt out (from the HX) = Primary Inlet

Coolant salt in = Secondary Inlet
Coolant salt out = Secondary Outlet
I am referring to the fuel salt as the "hot" fluid and the secondary salt as the "cold" fluid throughout my explanation and calculations. I apologize if this has caused any confusion, but I have always been able to keep it straight.

I suppose it may be better to refer to them as Tfuel,in or Tfuel,hot in the future.

mlippy38
Posts: 58
Joined: Feb 12, 2010 5:07 pm
Location: Blacksburg, VA

### Re: Heat exchanger

iain wrote:- Have you considered how the fluids get in and out of the channels? I looked at a slot design earlier, and found this to be a source of great complexity and flow restriction.
I have some interesting concepts that seem rather simple to me, but will ultimately be dependent on how many modules are used. I am thinking of it as basically an extension of the channels that move to a common plenum. This will result in higher pressure drops in areas that do not transfer heat, which is not optimal. I have a few simple models I can post some screenshots of in the near future.
iain wrote: Consider decreasing your channel length by 10x, and increasing your flow cross section by 10x. Same area. Note that the resistance to flow goes way, way down. Car radiators push this very var, and have channel lengths for the air stream which are a small fraction of the sqrt(flow cross section).
This is an interesting idea that I haven’t looked into yet. Meshing a design similar to that may be more of a challenge in FLUENT, but I will consider shorter lengths in my subsequent simulations.
iain wrote: I like your idea of putting the HX around the core and using it to route fluid from the top to the bottom of the core -- this minimizes out-of-core volume. Now consider what happens if you put the HX INSIDE the tank holding the core fluid. Ignore the irradiation problems for a moment. You end up with a tub of molten salt, with radiators around the periphery, slow up flow in the center and faster downflow between the radiators and the outer wall.
I thought long and hard about putting the HX inside the RX, but it just seemed to have too many complications for the few perks that it did give us. Maintenance would be among the chief reasons this would not be practical.
iain wrote: Now consider the radiation problem on the HX. How can you shield the HX?
Just add a greater thickness to the outside of each module to provide sufficient shielding.