Discussion on Group Activities

that well that will be our group discussion for today because while it's interactive you can discuss with us and if you have questions about the openmc part it's going to be in the discussion of tomorrow from 4 to 5 30. so today we stick to open Farm engine form and we do that till the end so go ahead to assignment three and in probably 20 minutes or so I will maybe half an hour I will give you the solution of assignment 3 and give you uh assignment for we'll start it and we will probably finish it tomorrow during the offbeat presentation because Alessandro told me it needs just an hour for offbeat so tomorrow we will finish assignment four so assignmentary today we start assignment for wishing finish assignment for tomorrow and openmc we discussed it tomorrow afternoon have we decided to do it tomorrow funny foreign do you have the key effective or is group one okay group two not yet group three zero point nine four one another digit another one group four foreign right this is interesting let's say so this one so you are grouped two right one which kind of Divergence game have you used girls upwind Carla you find the reason why I think they are both they are both correct okay I speak about six PCM difference I think the difference is that some of you started from the previous solution did not solve for fluid mechanics solved only for neutronics using the previous velocity field some of you started from the previous velocity but solved for couple neutronics and fluid dynamics is that the case so for those of you with the that one where is solving also for fluid dynamics or not only the neutronics no Fleet Dynamics for those of you obtained this in your control deck did you have solve three mechanics to true or to false true so that is the difference so what happened is some of you use the velocity field that we had before and solved on that some of you recalculated it and simply in 10 additional seconds it changed a little bit because before we we found our field after 10 seconds which was almost a steady state if you run it for another 10 seconds apparently changes a little bit so that's what happened for those of you who kept solving fluid dynamics you obtain the solution it is fluid dynamically a bit more converged so both are correct the difference is very small it's just about how accurate you are so for those of you who resold free Dynamics you just sold for additional 10 seconds you got the velocity field that is slightly different and that means that you have a slightly different K effect why you have different scale effective not because the velocity field impacts how you do how many precastors Decay either outside of the core or in low important low nutrients importance regions all right shall I do the exercise for everybody okay let's try to do the exercise for everybody I will try I said to start from assignment to write yeah too late which is the Delta activity stuff so this will be 1 minus Guided by the the map probably you will end up with how much yeah get something with 120 PCM which is in the order of magnitude of your beta because we are saying that we are moving the precautions so this has to deal with the uh your delayed Neutron and this is well I don't know I don't know which is the beta here do you remember Carlo um fraction here the effective static yeah approximately 300 300 so we are more or less in the same order of management that which makes totally uh totally sales sorry this is minus because we are we are losing so this means that our with respect to the previous case our Neutron delayed neutrons are created in zones that has low importance so so in which they contribute the probability to contribute to the uh Neutron reactions chain is lower all right um check what I do all right for those one who did it right check what I'm doing I'm getting tired and I may forget things so um we said we would start from case two so for the moment I just copy pasted case two into a case three and we said we want to start from a good guess and the good guess was the solution we had before so what we can do is we go into zero and we pretty much remove fluid region and we take it from time step 10 of the previous case so we just copy paste uh not sure why it didn't sorry huh not specified there are wonderful something oh I might have it open in not sure where actually apologies to interrupt briefly but we just checked for the posters and can you raise your hand if you did not yet put your poster there one two if you did not if you did not one two three because we are missing still I think about six but you can just get it already printed here at the info post but they will close in 10 minutes so you should just go ask them to to get it if you don't have it yet print it if you have it with you and you just need to put it up then it's fine so I'm removing Brute Force oh yeah no it did disappear this is what happens when you use WSL sometimes it does strange things so it would it would say in window that I deleted it that I couldn't delete it and now it's not there anymore so apparently it worked um WSL sometimes will do that so my suggestion is always if you can use directly Linux um so we go back to our starting case stream zero now we don't have fluid anymore take another wind off starting case three if you want inside 10 we still have the results we had before from fluid region you take it you copy paste and you have a good starting point for a fluid Dynamic solution and now what we want to do is solve the coupled problem so starting case system contradict and it's a little bit small I'm sorry about that but 0 10 0.1 everything we did before is more or less correct it depends on when you want whether you want to solve for fluid mechanics again or not according to Ali we can avoid that right let me try so let's imagine we don't solve for food mechanics so we will try to get his number so we leave the neutronics as it was 0 10 0.1 um actually anything else no I don't change anything in contradict let's try it let's see I will check that's the nice thing of simulations we are not using a real reactor will not break it it looks like let's see I did not so you guys did something different it's not yeah please uh what you mean no number iteration zero is because we are doing a steady state and you are implicit lighterating for 100 times so if you iterate even within the time steps you will be doing the same thing um I'm curious because in my own how I would have done it I would have done what I did now and solved the three Dynamics again getting what people got that did and some of you managed to get it without re-running with Dynamics and curious what you did comes in zero uh they took the end folder of case one that makes sense so it's true that you have two ways to get huh started from the end folder of the case one uh-huh you didn't call you didn't copy paste so you have two options here sorry for about the confusions you have two options either you did what I did now so you start from zero you copy paste your fluid but then you solve again for free Dynamics why because the neutronic solver needs to go get data from uh the fluid solver so you need to run them couples for the neutral region to realize oh I need to take the fluid um velocity if you don't do that it doesn't know it it doesn't get it so either you do what I did you start from velocity and just also solve for free Dynamics or I believe you guys started from folder 10 of the case one right this is where did you place it at the velocity you took the Velocity in the fluid of the velocity of the neutral region which is exactly what I did and we're getting we are not getting the same result so either I did something different or you did something different and we have to figure out who did something different this what I'm showing you is wrong so while we figure out what he did I think I know but we try to understand I will show you what I would do which is in this case we get we copy pasted the fluid solution into xero to get a good guess that my suggestion is solve it again to make sure that the thing is coupled even if it takes more time so what I would do is inside I would enter this the contradict so system contradict I will start from zero where we have our good guess and in 10 and now we are solving fluid dynamics again so you have to have a good safe initial time step and you will need to adjust the time step because we are doing fluid mechanics again so we do wait a sec I don't find that just time step I'm a bit blind In This Moment oh yeah I cannot hear you sorry I'm actually curious to do that I will try something sorry for the test just give me a sec I think we have to do what I'm doing oh Jen four more thing I will haven't forgot anything it will run for a while because we are solving again hmm oh you know what you can make it easier since we are starting from a good guess we don't need to resolve again for 10 seconds we can solve for one right because we are essentially having converged field we don't need to redo it again so we will just solve for one second and you're right I forgot the Phoebe scheme right of I think I have a better time copy pasting so this is what I was telling you um sometimes you will think get things that do not converge in open form and in pretty much any Library numerical Library out there sometimes you will have to touch the numerics and when things don't converge in fluid dynamics most of the time you get them to converge using an upwind scheme upwind scheme is relatively inaccurate you need fine meshes that is bounded oh you do that you same thing as for the other fields you tell the name of the operator that you want to discretize and you say instead of Gauss leaner because the default was Gauss linear you say chaos upwind and I believe it is correct I think I'm missing the underscore right give me a sec yeah absolutely so we are just telling um transform discretize the precars the Divergence in the precursors equation using an upwind scheme instead of a linear scheme those of you doesn't have any idea about what an upwind scheme is I can tell you actually so when we discretize the Divergence so imagine you have two cells what is the value of the field upwind is simply saying okay my value on the face is equal to the value in the cell that is happening to me so if you have a velocity like this you're just saying okay my value here is equal to this one plus that when you do linear you're saying okay my value here is the linear interpolation between my value here and my value here you can understand how linear is more accurate because we are saying okay my value is actually an integration between the two versus same well I use the one in the other cell but saying I use the one in the other cell is more stable um We call we call it a bounded scheme um why did I lose it oh yeah or Ctrl s close it so I think I changed everything I wanted to change let's see if it works so what we did we copy pasted our velocity field into zero to get a good guess and we are simply resolving from 0 to 10 both free Dynamics and neutronics to get a solution that couples the three Dynamics and the neutronics this is going to be slow so while we wait for this I would like to ask if you have any questions and then I will give you assignment 4 that we will finish tomorrow if you have any questions about this even slower than I hoped for but it's not that it is correct so when you discretize things uh you can have we speak about orders of convergence we speak about first order second order and essentially when you are first order instead of second order you have said to have the same accuracy you have to have a finer mesh so you are less accurate you're not wrong Your solution will always converge to the correct mathematical solution as far as you have a fine enough mesh it's just that it will require a finer mesh that you can see how it is less accurate I mean saying that the value here is equal to the value here it's an approximation saying that it is the average between here and here is the better approximation and the very general rule of thumb in numerics when it's more accurate most of the time is less stable at almost inevitably always happen and fluid dynamics is the bad guys on linear is complicated to solve Divergence is the most unstable operator to solve it will happen quite a few times that you have to have an upwind instead of linear scheme and neutronics we want both no we want to have a different so what we did before is we solve nutronics and we get a case for a static fuel now we are solving Us for free Dynamics so our field will be circulating circulating fluid means that our precursors will move so we want a couple solution we want to see how neutronics is affected by circulating fuel does that make sense am I not answering the question probably you can sure we have time you see how slow this thing is say again can you switch on your mic I I just wanted to like I just wanted to confirm partially for my sake uh that when we are bringing in the solutions from the previous exercises for the fluids we want to bring in the solution from exercise one but the neutronic solution should come from exercise two actually for the electronic solution you can take it it will not change much yeah it's also really cheap right compared to this it's very cheap so you can start from a zero flat flat not Zero from one flat flux okay um and it will not change much but you can you can initialize neutronic starting from exercise two fluid Dynamic starting from exercise one and use both The crucial part is fluid dynamics this is the one that takes long bread yeah I just want to clarify that where the solutions yeah yeah uh actually why I was speaking I said you could start from zero or one it's not the same thing if you try to start a solution with a flux zero I think you will get a floating Point exception because we are using power iteration and what we are doing is that we are normalizing our solution based on the integral in the previous solution and if your previous solution was Zero it will give you a miss an error so you have to start from a flux that is not zero when you solve for neutronics this is little something that you have to keep in mind unfortunately we do not have a check like if there is zero tell the user that it shouldn't be zero is something you have to figure out but I guess the moment you get this floating point you may think that there is some divide by zero somewhere um it requires some experience uh we gave I mean if you look at my slides this morning and the slides we gave on Monday we give rules of thumb some what resources you need how much it will take the how much resources you need it's easy um think about the 30 000 cells per core you know how many sets you have you know how many Korean needs memory depends if you're solving for fluid dynamics very often fields or for discrete ordinance where you have one thousand can be one gigabyte per um million cells so you can have several tenths of gigabytes per million cells it depends on how many fields you're solving for typically discrete ordinance where you might have 30 directions and 30 energy groups and all of a sudden you have 900 fields memory requirements are a lot in free Dynamics you have 10 Fields much less so it's it's you can calculate them you have to keep in mind these numbers about how much time it requires well it depends on how many time steps you are running if you have a solution that you have to run for like I'm working a little bit on nuclear thermal propulsion you have something that moves at several hundreds of meters per seconds you can imagine that the current number is horribly small you get down to 10 to the minus 6 and you have to run it for minutes so we have a lot of 10 steps you have 10 to the minus 6 time steps that you have to run for minutes so it's a lot it's going to take a long time if you do something like we are doing today we have a quarter number of we have a Time step of zero point something for 10 seconds we don't have many cells and it's gonna take probably 10 minutes so depending on the following you can go from minutes to hours to days there is no there are some rules that you can use but there is no General answer to the question the Darcy for shimer I'm not using Darcy for Shiner actually in these simulations say again it should be 2 power to 7 and you should get that by setting eigenvalue to true and by setting P Target in the reactor state to two to the seven Stefan is to add on the question with the time estimates what you can do is you can look at the log file and you will see there's a workload time and you can just take the difference between let's say a time step and the previous one so you know how much time the solver takes for one iteration or one time step and then you can make some estimates with the estimated number of time steps so you get roughly a time of the Run yeah all right so that's a way to estimate how long to wait that's true you can estimate at the beginning yeah wait you should wait until you It's tricky because imagine you have a simulation when you have a loss of flow your current number is going to change it should run seriously then it's okay if you keep let's say let's say adaptive time stepping and varying the flow rate of this won't work yeah but there are several ways and after a while it becomes an art to understand how long it will be your solution looks like it's going in the right direction so while we wait for this thing I would like to is there any questions otherwise I would like to at least give you an assignment that we will complete tomorrow and I want to give it to you so that you can start thinking about it um assignment 4 is the final steady state multi-physics we are going to add energy so the temperature solution um let me find it so the idea is we add the solution for energy we as usual we don't start from scratch we copy paste the previous folder and we want to solve for temperatures still the achievement of a steady state now to make it less computationally expensive I will ask you to assume that the fluid dynamics is not affected by the temperature which means we keep the velocity field that we already have is that always the case do you know if usually we can say the temperature does not affect our velocity field anyone how does temperature affect fluid dynamics which way if you parameterize viscosity well buoyancy there is another one another one density is going to change your uh Mass conservation equation is Gonna Change now we don't have a guess we have a liquid and we have an extremely high convection we're speaking about the reactor where the field moves at 2-3 meters per second so density is Not Gonna Change much buoyancy is not going to change our solution much because it is mainly Force convection so the reason why I'm telling you you can forget about how temperature affects fluid dynamics is because it's a very good approximation it's not perfect but it's a good approximation we have a liquid and we are enforced convection now the tricky part and I think I will need to give you some help tomorrow about this you will need a heat exchanger now in gen form there are numbers of way to simulate a heat exchanger which makes things a bit more complicated there is one that is easy and that is the one that we very often use when we do multiphysics we assume that on the secondary side there is a fixed temperature and what I would like to have is a secondary side with this um fixed temperature of 900 Kelvin you will need the volumetric area do you know what a volumetric area is is an area of heat transfer divided by volume all right so if you have I don't know 8x6 yeah all right so imagine you have a very strange heat exchanger where you have primary fluid and the secondary fluid and you have one tube it's the worst heat exchanger ever your volumetric area is the surface of the tube divided by the volume of the heat exchanger is giving you in um intensive way instead of extensive way what is the surface area of your heat exchanger it's just this area divided by the volume is something you need you can understand this is something you need I mean heat exchanger is characterized by a volumetric area so you will need to give it how you will find it out um the tricky part about this exercise is that if you were learning gen form in time having a few days you would look into the options you would read the documentation and you would find out to set a fixed temperature something somewhere in the frame of an exercise like we're doing today and tomorrow you don't have time to you know look into the documentation so I'm telling you and I will tell you tomorrow when we do it the power model that you need to use is called literally fixed temperature all together fixed temperature I'll tell you again this tomorrow but you can understand that if you want if ever reactor that now produce power you need to evacuate the power if you need to evaluate the power you need a heat exchanger the easy way to set it up is okay I assume that there is a subscale structure that in our case is the fluid in the secondary circuit that has a fixed temperature of 900 Degrees this is our heat sink so this is what I would like to do tomorrow we will take an hour in the offbeat presentation to try to finish exercise four why I want to finish exercise 4 is because this will give us a fully multi-physics solution where we have temperature neutronics and movement of delay Neutron precursors um and this is a solution of a fairly real system with the real 2D geometry starting from there it's gonna be easy to run a transient all right so what we will ask you tomorrow if you can after the course try to stay here go home go in your room go with your group somewhere and try to run a transient but I will show you tomorrow that will be assignment five it's you will see it's a relatively easy assignment compared to what we did today but step by step today I will show you the solution that we got for assignment 3. tomorrow we'll do assignment for and we'll give you as assignment for Friday to do one transient now let's hope that our simulation has converged hopefully almost are we far from that or one nine four one nine four one nine four one nine four one is that the thing that's for one nine four one seriously um I clearly had a problem so I will give you the solution tomorrow I think I forgot to set the parameter somewhere huh again oh thank you I'm in the wrong common or maybe explain the thing oh yeah you're right nine four one nine two four and four one nine two four man I'm slightly different likely because I sold for one second instead of 10 seconds again depending on how long you do it you change a little bit the solution two four we are two PCM away from that solution so it probably because I run it for one second a new guy ran it for 10 I assume but it's okay so I didn't do a mistake which is good and we can look at the solution so first of all well the key effect you find it there you find it in many different places as Stefano says it's smaller than before because we are losing some of the precastors outside of our reactor we look at the solution after some time so we ran to one so we have to go to time step number one refresh let's see if it worked so we looked at like fluxes let me just look at the neutral region so otherwise it gets funny um did you expect the flux to change with the Newton flux to change with the movement of the neutrons no anyone will expect it to change think about it our neutrons are moving at what's the speed of a thermal Neutron more or less order of magnitude 2000 we are in a fast Spectrum reactor what the speed of our neutrons a lot doesn't matter a lot even if it was thermal we have to take about 2000 meters per second our fluid is moving at two so the fact that our nuclei are moving compared to our reference system by 2 out of 2000 which will probably be two out of 200 000. does it change our solution no usually when we solve for when we study molten salt reactors we assume that the neutron field is not affected by the velocity of the fluid which is typically a nearly perfect approximation it's even hard to call it an approximation things change when you want to simulate when you want to um stimulate precursors as gets number seven so number seven is a relatively fast decaying precursors and even though it's fast decaying we still see the effect of Transport so our flux was perfectly centered here you remember now our precursor for Group of Seven which is the fastest Decay are slightly above what's happening is that our precastors are born in a place that they Decay somewhere else and their pick is actually moved somewhere else so our precursors are moving if you get something that is moving at this decaying a bit slower like group five you'll see that effect even more why because they are Decay more slowly they have more time to get away from the center of the core and you look at precursor five I think it's a good example because sometimes we tend to think that the only way we lose precastors is that they go into the heat exchange or the pumps and they are outside of the core but that it's an extreme case what's also happening is that our precursors globally are moving to low flux region now if you're a bit Neutron transport Savvy you know that flux and a joint flux they tend to be very similar so you are also moving to regions of your core where that joint flux or the Neutron's importance is smaller so the value of your precursor decaying here or here is not the same so you are we are losing bit effective also because some of our precursors are decaying in low importance regions of the core so this is something that is extremely important in molten salt reactors so let's forget a little bit about genfoam let's speak about physics in this reactor if you have a static field you would have a bit effect if that is already small because this one is a thorium-based reactor the uranium plutonium version of it let's say the Terra power version of it is uh uranium 5 but imagine you want to use plutonium because you want to do burn so both for tutorium and for plutonium already your beta effect if your static bit effective is 300 PCM more or less do you know how much it is in a pwr 7 800 so we already have a bit effective that is small and all of a sudden we lose part of it because you lose delay Neutron precursors outside of your core and you don't lose a little bit of it in this half of it your bit effective goes from 300 to 150. all of a sudden you ever reactor doesn't have much margin to prompt criticality this can have significant impacts on the design for instance you will not typical thing that we do when we design a reactor we say okay we don't want our control rows to exceed bit effective why because you don't want be in a situation where you extract your control Road and your reactor is super prompt critical so for instance that means that in a fast reactor like this one your control rod will have to be limited to maximum 100 and something PCM you want to achieve a cold shutdown of this reactor and you want each of your control row to be maximum 100 PCM you're gonna need a lot of control rods and this is problematic for this reactor this is one of the reasons why people think about okay we don't get to a cold shutdown we get to a hot critical state or we change the shutdown mode we dump our soul to in critically safe dump tanks so we empty our reactor we put the fuel into tanks that are critically safe one of the reason is the small beta and actually the very high feedback coefficient this reactors fit the coefficient on the order of almio Stefano 7 PCM per Kelvin density plus temperature five and seven PCM Perkins a lot on minus negative which means if you have to go from 700 Kelvin to room temperature 300 is 400 kelvins multiplied by seven it's too difficult three thousand more or less almost three thousand PCM you would need three thousand hundred and fifty twenty twenty control rows authentic controls is a lot in this reactor and also controls will change your view other problem is that you don't want to put controls inside the reactor why because that will affect your velocity in the reactor you don't want to mess with the velocity so very often when we speak about molten salt reactor people think about using control drums pentecontrol drums that get 250 PCM not really so it's very complicated again they're very ineffective so it's the whole thing of the small bit effective it makes the old design of the reactor a bit complicated I just wanted to conclude the demo day on a more nuclear engineering oriented note um and you know also to mention you know you can use how you can use these tools to get data that are essential to design that are essential to understanding a reactor and I'm pretty happy that we got here in one day so congratulations thanks for trying to get there tomorrow we will try in an hour or so to add temperature solution to this all right um how are we in the program okay should we do poster and just because we have we calculated nine only positively so I think we can do um to do it today and also into the old all the posters okay okay then we can do it today okay so we have time okay so we can do the post now straight away or we give 10 minutes right straight away slowly slowly okay let's slowly move to the poster please try to prepare yourself and again just to remind you because we are it's just three minutes presentation and then few minutes answering questions um wishing you all good luck today foreign [Music] how will power maneuvering be done in for such reactor this is a very good question probably Stefano can answer but I believe the strut there are few strategies that you can use you can use variable speed pumps you can drive um using the secondary circuit and you can do control rows you can it's not forbidden I Am pro controllers but the thing is you can have controls for control or for safety shutdown safety shutdown is more complicated because you have to compensate for 3000 PCM yeah but we'll be mailing you with the power extraction which is actually what you do in in light Water Reactor control rod are not used to control your power controlled in light Water Reactor are used to control the temperature the average temperature in the light Water Reactor here probably we don't need to control the average temperature even if because if you just you are just extracting power the average temperature will be constant because you are not adding any other reactivity coefficient so the for the power maneuvering will be mainly do just through extraction you are extracting from the uh intermediate or the uh Energy commercial system less power but control order mainly for safety reason rather than let's say control reason it's called safety rules [Music] I'm not sure I understand the question um if you mean calculating the bottom concentration that will make your reactor critical we don't we use an external reactivity which is equivalent to using Boron we can use Boron concentration as an external input to the point kinetic solver to as an additional feed so if that answers your question otherwise please ask again be more specific about what you mean by calculating Boron concentration all right then thank you thank you Carl thank you all and thank you online participants and we will see you tomorrow at 9 00 am