Caedium free trial feedback

I have spent one day testing your software Caedium Professional. I have mainly concentrated on using the flow solver since I already have a mech design software that I like.

I have found a number of things that you need to fix before your software is usable:

1. Don't use ridiculous sounds when I select things in the menu or run simulations. Its not a game or a toy i presume.

2. The graphics look okay (it's not ugly), but the handling of zoom, pan and especially spin is really stupid. Have a look at any modern mechanic development tools where you can do all three with simple mouse gestures. Use the middle button for spin when pressed, zoom with scroll wheel, pan with right button. Left buttons selects objects. This will also prevent you from other (silly) usages of right button.

3. There are far too many, right click - left click - right click, for selecting things. This is just plain stupid and hard to remember. Make things more consistent. Preferably object oriented.

4. It is impossible to know what has been done, selected or modified in the model. There is no visual indication or feedback that a surface has been selected as input or output (or any other things). I can't see anywhere that it has been done and I don't find anywhere to change the input or add a second one or remove the input. You could fix this by having a tree like structure like other mechanical design software.

5. I cant find any easy way to alter the boundary conditions. It seems plain stupid to me that I need to select flowing media (air) ones again to be able to change the velocity. Besides this the velocity has got quite little to do with the media. That feels more like the input. What if I have two inputs? What do I do then?

6. The graphics is far too slow. Use a better 3D engine. Take something from the gaming industry. My regular laptop almost stopped completely when doing the tutorial with the simple pipe-into-box.

7. Get things straight in you tutorials! There are no "Scale"-field in the properties when I have displayed the velocity vectors. The file I downloaded with the pipe-into-box has a scale factor error of 1000 or the software has a scale factor error of 0.001. I needed to translate the surface for the particle tracks 3500m (not 3.5m as the tutorial say). Probably as a result of this the calculations in the end doesn't converge. This really sucks and makes the tool or tutorial (can't say) useless. There are no way as I see it to measure distances in the model. That sucks.

8. I imported an IGES file that I exported from my mech design software. I took forever and consumed all memory of my computer and before Caedium finished I had to kill the application and reboot my PC to get it straight. This IGES-model is not extremely complicated, but not trivial as your tutorial example.

9. Your user interface is very far from intuitive. Perhaps you should try a more object like interface where you select either one or many edges, faces or volume by left mouse click or left down-drag window-release move. Either in view window or in a tree like structure mentioned earlier. Have a look at any modern mech design software, they all look quite the same.

10. I can't seem to figure out how to stop and restart a calculation. Stopping went fine, but upon restart it seemed to continue where it stopped even though I changed some input parameters like accuracy. Even when I saved the work, exited Caedium, restarted and reloaded the workfile it still continued where it left when I stopped it. I didn't found any restart button. I suggest you totally disconnect the simulation-run, -stop of calculations from your recording function. It's not logical.

[Edit: based on an email exchange]

Not Using Latest Version of Caedium

Sorry to hear you didn't enjoy using Caedium. I think one of the problems you are facing is using our legacy 32-bit version rather than the latest 64-bit version. If you are serious about performing CFD then you need to consider using the latest 64-bit version to have access to the larger memory address space. To see some of the functionality and improvements you are missing (v2.2.0 onwards) review the Caedium release notes.

Specifically in answer to your questions:

1) If you'd prefer to switch off the sound feedback then set the preference accordingly, File->Preferences, then in the Properties Panel turn off the Application->Sound preference.

2 + 3) The mouse buttons/selection are as is for the moment, I'll consider your suggestions further, thanks. By the way everything about the Caedium interface is object oriented - volumes, faces, edges, vertices, tools etc. are all objects with their own properties. If you make a group (multiple) selection then you can make property changes across the entire group.

4) When you select a geometry entity the edges are highlighted if it's any help. If you want a model tree, then select the background and the Select dialog has the entire model in a tree.

5) To change boundary conditions (BC) you select the entity (e.g., face) with the BC, then in the Properties Panel select the Entity tab (e.g., face) then make changes to the Physics tree properties. You can make explicit changes to any number of BCs if you don't want to refer to the reference values.

6) The latest version of Caedium has improved graphics performance.

7) I think your tutorial problems are caused by you using an out of date version of Caedium.

8) Larger IGES files can be a problem, no doubt. If you have the option I'd suggest using STEP, it preserves the topology much better than IGES. I suggest also if you can, try reducing the model in your CAD system to the essential features you want to simulate flow around. Usually it's much easier to de-feature a model in the native CAD system then in Caedium after translating the model.

9) The Caedium interface is object oriented, more so than most I would claim. I think your problems/criticisms relate more to how you select things. See 2 + 3).

10) I'm guessing that you are using the Stop button rather than the Run/Pause button to start/stop the simulation, for more details try: "Run, Stop, and Lock Controls".

You might also find our screencasts useful.

Are you familiar with any other CFD applications, such as FLUENT or STAR-CD?
In comparison I think you would find Caedium much more intuitive, though clearly we have a lot more work to do. Of course I'm likely to be biased too :-)

Thank you for your answer

Thank you for your answer. OK. I realize that I have tested an old version. I only have access to my old laptop with windows XP (32 bit) for the moment. I will have access to my new 64 bit machine in a week or two. Perhaps I will perform a new test with that

I don't have any earlier experience with CFD on my own. I am an electrical engineer (MScEE) with skills in mechanical engineering and embedded software design. I design (mostly) electronics as a hierd engineer. My clients have asked for simulations of air flow and heat exchange and I thought it was good to test this for myself. I am using a lot of simulations in my designs (electrical simulations with SPICE and mech simulations (forces only) with COSMOSExpress built into SolidWorks). I would like to be able to recommend my clients a good CFD for heat transfer. I will test other CFD/FEM simulators and attend seminars in the close future. I like your idea "CFD for all". That is the reason for my tests and I see the feedback to you as my way to give something back and hopefully get a better application in the future. I hope you don't find my email too long or uninteresting

I continued with Caedium using STEP-files instead of IGES after your recommendation. That worked better, but I couldn't get Caedium to handle volumes assembled from multiple objects. I also didn't manage to simulate flow in voids (volumes trapped by solid (thick) walls), I had to convert the voids to volumes and remove the wall thickness. This will be quite hard with a model of a circuit board

I also couldn't get Caedium to solve the equations when I have a slanted wall for input or output. I am not really sure that this is the reason, but I performed several tests that show this. I got results that say that the flow of air makes a narrow right angle turn when it comes into a large open volume. I attach the simulation file so you can see what I mean. The geometries are just test geometries so I have no real data to compare with, but the result is just ridiculous. I don’t see why the flow shouldn't be close to symmetrical in a completely symmetrical volume. I could have done something wrong though. Please let me know if you want more information like the STEP-file or native model file or if you find whats wrong

I think a lot of my frustration with the controls is that I have a feeling that I need to answer "yes" to the same question many times for everything that happens. Like everything needs to be confirmed twice in two different ways. I think that selecting things should be one mouse click only (left!). No confirmation. Then a clearly visible highlight of that object (not just thicker black that can’t be seen when out zoomed). If my select is wrong then I just click ones more. If I want to apply a command I don't want to confirm that or press apply a second time. If it goes wrong then I "undo" that command and simply change the feature of the command. (Ctrl-z should preferably be implemented as undo.) An applied command needs to change something visually. I.e. no "silent" commands. For instance if I add air in a volume this could me marked with (x) after substance or something

Besides that I don't think that the media (air) got so much to do with the speed of the air. This should be set separately. It feels ambiguous that you can set the air speed in "air" and then there is a "velocity" command that does (?) nothing. I couldn't get it to work. A vector is nice, but setting a scalar value for velocity that is perpendicular to the input face is probably much more common. I didn't find anywhere where I could specify a (back) pressure as a boundary condition in the outlet. This could be due to inexperience from my side, but its not easy to find. It could also be useful to have an option of setting the output flow to be laminar. In the example I send you I have a reverse flow velocity in the outlet (input flow that in the outlet) in one point of a magnitude more than my inlet flow!&newline;I suggest you use colors in your wire frame/solids models. Like green for things selected, red for input surface, blue for output, lighter gray for faces that are not walls. A simple button to switch between solid walls and wire frame would be nice. A simple button to switch on and of the mesh. Why do I need to select all faces as walls? This could be default. The fluid inside the volume could also default to air or whatever selected last time. Particle tracks and vector fields can be done much easier like a pop up dialogue when right click a face. You could use left button to select and right button to show preferences. WITHOUT asking for confirmation! Ctrl+left button selects more (or removes selective). Pushing several times in the same location toggles selection between the possible things to select. Edges are selected on edges and faces only between edges. Volumes should probably never need to be selected in the view. They can be changed in an operations tree that should contain all starting conditions clearly viewable/changeable. Colors of highlight and also mechanical parts like wireframes should be selectable or visual or not with a simple button. (All "simple" button suggestions refer to buttons in a toolbar.)&newline;I read through the "Play, Interrupt, and Freeze Controls". This doesn't make sense to me at all. I either like to change the simulation conditions like boundary conditions, change my model, speed and start all over with the simulation. Or I like to save the result. I can't see that there is anything between that I like to "roll back" to. These buttons seem to destroy both my boundary conditions and the unusable result and in the same time not resetting the software enough so my next result will be impaired by the previous result

I think that a good idea is to make the simulation able to run in two separate stages. Like one button for meshing and another button for differential equation solving. It seems like the mesh is viewable when the faces are solid (non transparent). The mesh could also be viewable in wire frame mode (transparent) with a separate toggle on/off button. As it is now the meshing takes place by itself when accuracy is changed. Say that I have a simulation to do over night, then I like to wait with the mesh until everything else is ready to begin

Don't put any more effort in the mechanical design parts of the program. You will never be able to come close to a mech design software. Its better to put effort in import/export of model data

I didn't understand this part of your answer: "If you want a model tree, then select the background and the Select dialog has the entire model in a tree." How do I "select the background"? Left click in view window does nothing, right click gives a dialogue for model entities, but it contains no information on the fluid media selected (attached to volumes), no information on boundary conditions, no marking on input surface or output surface, nowhere where I can change accuracy or see what is selected. The mechanical entities are only a very small part of the starting conditions. Please refer to the design tree of SolidWorks (similar to many other modern mech design SW). I think this would be a better thing to have on the screen instead of all the commands. Or why not list the selected faces under "inlet"? Then I can review and change the selection. Usage of +/- like an expandable tree helps

As I wrote before, no I haven't used any other CFD but I don't see a reason to why this should be complicated. I know that the FEM-analysis is not trivial, but the user interface could be. It feels like a poor excuse that there are others that are worse.

Thankyou for more Feedback

Thanks for sharing more feedback on Caedium.

I took a look at your project file - it seems your inlet and outlet faces are under resolved (mesh is coarse). By using the Accuracy tool on those faces I was able to resolve that problem. The other problem is that you are initializing your flow with your inlet velocity, but your inlet opens up into a large area, so I used 5m/s instead (Physics::Initial->U:Fixed Value->Value = 5).

I noticed that your geometry is massive 330 x 400 x 150 metres - was that your intent?

I think you'll have a much better experience with the newer 64-bit version of Caedium, at least in conformance to the tutorials anyway.

You have to supply/create a single flow volume. I'm not sure what you are expecting Caedium to do with multiple volumes?

Thick walls have to be part of the fluid region i.e., subtract them to form a single volume.

In the 2.1.0 (32-bit) version you have, there is enhanced meta-key picking, see the Selection section in the release notes.

"An applied command needs to change something visually. I.e. no "silent" commands. For instance if I add air in a volume this could me marked with (x) after substance or something."
There is feedback when you DND air onto a volume.The Properties Panel switches to the Volume tab and Air is shown expanded. In addition it's highlighted in pink, which signifies you can delete it.

Ctrl-z is implemented as undo.

You are setting a reference velocity for the simulation when you specify the velocity on air. It is used to calculate results etc. and as a convenience the boundary conditions (e.g., Inlet) and initial conditions can refer back to it. This allows you to change the reference velocity and have it propagate to the BCs and initial conditions. If you prefer you can visit each inlet and set the velocity explicitly - same goes for the initial conditions. The Linear Velocity tool you refer to is for specifying a moving wall, e.g., a moving ground pane.

Basic tooltips are available for most tools and properties by selecting them and viewing the status bar in the bottom left of the application.

After you DND Outlet onto your geometry the Properties Panel shows you the Physics:Outlet property. To adjust the pressure set Physics:Outlet->Type:Default->p/rho = Pressure and enter your gauge pressure.

I don't see any physical meaning to setting the output flow to laminar. Maybe you mean the entire flow should be laminar? In which case select the volume and in the Properties Panel select the Volume tab. Then set Substance:Air->Solver:RANS Flow->Turbulence = Laminar.

CFD is not linear stress analysis where the linear-solver either converges or it doesn't. There are times (quite often in fact) that the CFD simulation will not converge or might 'blow up'. In these cases it is often useful to take the simulation back to a 'good' state using rollback, make a change and then resume. I don't understand what you mean by "destroy ... my boundary conditions" - they are applied throughout the simulation - you can't destroy them. Did you understand the difference between Interrupt and Play/Pause? Interrupt is like an emergency stop - it will corrupt your simulation state.

You can create the mesh in a separate stage, and I often do for the reason you cited.

In my previous reply I wasn't apologizing for Caedium in any way. I think it is the best CFD simulation environment available today. However, there is always room for improvement and I am open to feedback. So please don't view my previous response in the vein "... a poor excuse that there are others that are worse." That was not my intent.

In my experience CFD is not well served by a 'trivial GUI' - invariably you encounter problems at various points in the process. Sure you can automate the whole thing, but only for trivial problems with trivial geometry and BCs, e.g., flow through a box. But where's the fun in that? :-)

Another Look

I just took another look at your original Caedium project file and even though the inlet and outlet faces are under resolved, the only change you need to make on your original simulation is to set the initial velocity to [0.5 0 0] m/s and it converges within 200 iterations.

Scale Factor Problem

Thank you so much for looking into my problems!! (It’s Saturday) I have now managed to get it to converge too. I didn't know that CFD was so depending on initial flow values to converge

I am looking into the scale factor problem. I saw this in my initial tests as well. Even with your tutorial I got things a factor 1000 too large. My model is designed to be 330 x 400 x 150 millimeters, not meters. I have exported and imported data to and from SolidWorks to/from several manufacturers and also machined things myself (CNC milled) without scale problems. I am now trying to find out the properties of the STEP format. I could convert all my measurements to micrometers, but that will not solve the real issue. And the end result is not that important for this test. I would just like things to work. Could this be an effect of unit conversions? It seems to me that you are using metric units everywhere already.

I can see numbers in my STEP-file to be in the size of 100.0 where SolidWorks measure my model to be 100.0 mm. I found some information regarding the step format, but the standard (ISO 10303) is not freely available.

I have now also managed to change the initial velocity in my BC. It took a long time to find where I could change this.

I know that CFD is quite a bit more complex than linear stress analysis. I know that turbulence usually doesn't only give me a quasi-stationary vortex (spinning in the same way forever). It will give me several vortices traveling away from the edge or object that will grow in size. I have never seen this in a CFD result. I have only seen quasi-stationary vortices.

It felt to me that when I pushed the "back to beginning"-button things got lost like improved accuracy. I realize the difference between play/pause and interrupt, but I don't see the meaning of saving states yet. Without going to beginning I can’t push “play”. If I am going to change things in the model or BC I don't have much use of old simulation results. Or do I? I can see that it might converge in some less iterations, but I guess only marginally?

I am sorry for using your old 32-bit version and consuming your time with old bugs, but I am very pleased for your efforts. I hope to get this running on my newer machine soon.

Good to hear you are making progress

Good to hear you are making progress.

I wouldn't worry about the STEP import scaling. In older versions of Caedium (pre 2.4.2) the scaling wasn't imported correctly. The workaround was to scale your geometry in Caedium - probably worth setting up an axis at the origin and scaling relative to that, for a scaling example search for Transforms->Scale in the tutorial "Flow Over an Airfoil".

There is nothing inherently turbulent about large scale vortices or recirculations. They can occur at any Reynolds number (laminar or turbulent). If you wanted to simulate the alternating vortex shedding say from the rear of a cylinder, you'd need to run your CFD solver in Transient (unsteady) mode. For an example of a transient simulation see "Incompressible Transient Flow with Heat Transfer".

It is always preferable to run steady simulations, as they will converge faster and you'll have less results to manage. However, this is not always feasible, say if your simulation is inherently unsteady then you will be forced to run in transient mode.

As background the Reynolds-Averaged Navier-Stokes (RANS) equations we solve in Caedium RANS Flow solves for the mean flow variables (e.g., velocity), not the exact near-random variations around the mean that you find in the complete Navier-Stokes equations. In RANS you can capture separations and vortices, but no small scale turbulence. Turbulence models (e.g., K-E) provide the idealized (to a greater or lesser extent - turbulence models are a weakness and a strength) missing fluctuation influences. The turbulent variations are significantly smaller than the mesh elements - this is how we can solve RANS for realistic flows in a reasonable time. If we had to resolve the small turbulent variations we'd need many orders of magnitude smaller elements - that process is called Direct Numerical Simulation (DNS).

Pressing the Beginning button resets your simulation, i.e., re-initializes it. What you see on your model is the initial velocity, pressure, etc. results.

I'm not sure what you mean by "saving states," but if you are only making relatively small changes to your BCs or reference values then it is worth resuming a simulation rather than resetting it The closer your current simulation results are to the final results the faster it will converge. In some cases the speedup can be significant - could be a factor of 2 or more.

To increase the range of the Play period (i.e., increase the total number of iterations), select the background simulation, in the Properties Panel increase the Time->Duration property. For more on the influence of Duration and Solver->Iterations, review the 'Run the Flow Solver' section in the tutorial "Transonic Flow Over the NACA 0012 Airfoil".

Note if you press Pause (simulation is running) then the simulation will stop at the next results update in a 'good state.' It doesn't hurt to have a long Duration if you are willing to manually press Pause when you are happy with the convergence. While iterating the sound feedback provides a sense of whether the residuals are converging or diverging - there are different sounds for each condition. There is no guarantee that your simulation will converge smoothly.

I agree completely with this

I agree completely with this users comments. The interface is horrible! A user needs absolute confirmantion that the surfaces he intends to select for grouping have been selected, or deselected. I'm a long time user of Fluent and Star CCM+, and both of these packages have beautiful interfaces. I realize this you get what you pay for, but having to guess or hope that a group of surfaces has been selected makes the software almost unusable. Having the edges highlighted is very hard to see and interpret. How hard can it be to have the selected surfaces change color when selected? This software is not ready to be releaseed as a commercial product yet. Let me know when you get serious and we might consider purchasing a license.


"The interface is horrible"

Is it just the selection highlighting you don't like or is there more?

FLUENT, "beautiful interface", really?

Boundary selection

I recently downloaded the trial and am trying to get acclimated to the RANS solver. I have used several other CFD solvers (GASP, CRUNCH, ADPAC), and I am intrigued by Caedium as a more accessible version of OpenFOAM.

As a previous poster said, one of the biggest frustrations is the boundary selection. I do not like doing this in the view window; it is much preferable to just select the boundary from a list format. Right now, to go back and check all your boundary conditions requires clicking all over the view window. A single tree with all the BCs is much easier to work with.

Other than that, I'm fairly impressed with Caedium and we may purchase a license, but first I'll need to get a converged solution for a current project. I have a 2-D airfoil with a very blunt leading edge operating at transonic conditions, but the solution always crashes, going hypersonic upstream of the leading edge.

Boundary Condition Filter

To select a specific boundary condition (BC) type on multiple faces try using the Selection Filters->Primary Fluid Conditions on the Results Tool Palette according to:

  • Select the Primary Fluid Conditions tool
  • In the Properties Panel set the Condition property to your desired BC type (e.g., Wall)
  • Drag and drop the tool onto the background of the view windows and in the Select dialog select sim->Faces

Assuming you've extruded your 2D airfoil to make it 3D then the setup and running of such a case should be similar to the tutorial "Transonic Flow Over the NACA 0012 Airfoil". However, the tutorial uses a special one cell thick mesh, assuming you don't have such a mesh you will need to assign Conditions->Symmetry Local from the Physics Tool Palette to your end faces.

Note that the transonic RANS solver assume that your freestream flow in subsonic.

Thanks for the quick reply. I

Thanks for the quick reply. I used all of the same solver parameters as the transonic airfoil tutorial you mentioned, and I did extrude my mesh to be 1-cell thick (the mesh was made in Pointwise). However, I still can't get this simulation to do anything other than go Mach 100+ in the region just upstream of the leading edge and then crash.

The state is set to Transonic, with a freestream Mach number of 0.625 (from XFOIL, I expect the flow to go supersonic locally as it turns around the blunt leading edge). I've tried all of the Divergence schemes, set the relaxation values super low, tried inviscid/laminar/several Tu models, tried various gradient limiters, all to no avail.

Share the case

If you'd like me to take a look then use our contact form to share your email address.