17KB
First of all, I want to thank you for this email for several reasons:
- you took away a huge amount of time for a thorough read and for writing these detail feedback. This has already payed off my effort!
- you indeed read between lines and grasped the “behind-the-lines” message. I am glad you did.
- you respectfully said what you think even though someone might take it badly. Luckily it is not my case, so this is a win-win because you told me I suck with a lot of sound arguments, and I like being told I suck (with sound arguments)
You are right and have a point in almost any of the issues you describe. If you allow me, I will try to respectfully argue back some of your points. Please Angus and Nick feel free to write anything you want. Probably Nick, who already read the case did not dare to write something like this but now he has te #meetoo hash :-)
On Tue, 2019-09-10 at 12:02 +0200, Łukasz Skotny wrote: Hey Guys,
I had some time so I gave a Jeremys CS a read.
My comments are below. Again I apologize Jeremy if I may sound harsh. As you know time is of an essence as always so I might use some shortcuts etc. which may make me sound inpolite or something. I apologize in advance. In the end the goal is only to have the best book ever, and this is what I had in mind.
You are not harsh at all. I loved everything you wrote.
I really like the beginning – when you start with the students “millennials” and all – it’s pretty cool and set up a nice “mood for reading it” – cool stuff!
The pendulum example was fun… you know something simple and then complex – I see what you try to do here
I don’t like you using me personally as an example, since it’s your text and I cannot make an argument to you. I think you really “flatten” my stance on math and make it look shallow and then at least several times you write something like “don’t fear math” etc. And… the math you mention isn’t even what I have in mind (a stress tensor for crying out loud – that is borderline insulting me). I don’t have the mental capacity to argue with you on this – I did it countless times already with various PhD and experts around the world, and so far only one guy proved to me that understanding complex math (as I understand it, so definitely not the stress tensor!) has it uses… and it was only because he modified his own solver to compute a problem faster (kudos, I would never achieve such a thing, but this is not a game changer for me). So please do not use my name, or even reference me or suggest that in any form in your text. Otherwise I will have to insist to use my arguments in full not the shallow version of “Łukasz is afraid of math, silly him” because I don’t fancy that
I get your point, because I wanted to say A, I wrote B (especially in a non-native language) and people understand from C to Z. I apologize if I insulted you, of course I did not want to do that. I will try to ellaborate.
As we all four know, we can use this text more like a way of reaching other people and eventually selling them consulting, training or whatever we do for a living. So when I first wrote it, I thought about trying to make a point about this issue of “maths” or “no-maths.” I have interacted with a lot of people with a wide variety of backgrounds and my life (both professional and commercial) would have been far better if most of these people talked “the same language”, i.e. maths. In fact the best projects turn out to be with those clients who understand and appreciate the maths behind the solution.
It is in this regard that I wanted to use the text for future reference. Of course you are not afraid of stress tensors, but you know most people do. And of course you have a PhD (I do not) and know how to solve differential equations, but many people do not even undertand why they have to set displacement boundary conditions in FEM problems. Let alone wanting to obtain “thermal stresses” by just solving heat conduction and not knowing what displacement boundary conditions they need. The spirit of my writing was this.
On the other hand, I do understand that you need to say to your potential clients “you do not need to know maths to use FEM,” and you are 100% right! I tried to make a point with the example of the driver who do not need to know thermodynamics, but I clearly failed. You see, if this text is going to be read by students which are more likely to know thermodynamics when driving a car, they already might know all of the math I present. For them (and for all of us) the math in the text in pretty basic, yet the majority of canned FEM software users will not be comfortable with a line integral to compute the bending stress. So it is a way to reinforce the students’ self esteem: hey, this guy says that most people fear maths and I do not, so I would have to be something!
I will hapily either
a. remove your name from the quotation, or b. explain better my point and your case
Again, the readers of your blog and the people that take your courses are plain drivers. The readers of the CSs are the team engineers of the race car.
I really got worried reading your text just above chapter 5.2.1 where you write “this is not a case-study but general ideas….”. This is not what we were hired to do I’m afraid. We agreed to demonstrate an industrial case-study of an actually solved problem, and I haven’t found that in your work, which was really disappointing. Also I can imagine Nafems may share this concern, especially since you even state yourself that this is not a case-study…
Sure, if I did then I would have to write a lot of sensible information regarding a particular nuclear power plant. This industry is very jealous regarding the sharing of information. I had to ask for permission to the plant owner, so I had to remove all numerical data and give only general information. They even accepted because I told them that this was primarily for students.
I will re-write that paragraph to reinforce this point.
Don’t get me wrong, your style of writing is really fun (to the degree that I was actually surprised, since you seemed rigid when you were commenting my style), I like know you have those small stories and “breaks” molded into the text – It’s a wonderful introduction to any topic. But this is also the problem as well. After 15 or so pages (near seismic loads if I recall) I was starting to feel a bit “tired”. I felt that I’m still reading the introduction, but I wanted to learn some practical things instead
Yes, I do not expect someone to read the whole thing in a single run.
(that is a purpose of a case-study after all). And so:
5.3.2.1 you are using several equations there, but I honestly have no idea what is the purpose of that. They just appear there without derivation nor explanation and I have no idea what is the point. The entire section can be summarized as what you write under those equations in 4 points, and that text alone would be just as believable, since in my opinion those equations don’t bring anything to the game.
The derivation should be in a referenced paper (I will check it is properly linked): https://seamplex.com/docs/pipe-linearized/
The point of those equations are that in school you learn how to solve this idealized case, and the result is just that: an equation. No other application whatsoever. Yet, you see that displacements are linear with pressure, and that stuff. This should be referenced back in the real case, noting that displacement will mostly be linear with respect to the pressure.
Will reinforce the idea because it clearly was not delivered.
5.4.1 I fail to understand how discussion about an industrial case-study included a topic about finite differences, volumes and elements. I mean sure, this is a pretty interesting topic but for a lecture set on FEA… not a case study
Ok, maybe you are right. I wrote this because more than 90% of the people I talk to (and it is 90% because I talk to a lot of people in the academia) think that FEM means “elastic problem.” And I wanted to illustrate that you can
a. use FEM for other problems, like electrodynamics b. solve elasticity with other methods like FVM or FDM
But if somebody else thinks it should be removed, I will remove it.
Then the kinds of finite elements comes – exactly the same approach on my end – I couldn’t wait for the case study to start…
It starts half a page afterwards! :-)
I really like how you explain stuff when it comes to pipelines and thermal stuff and all. The problem I have is that those are very “up high” explanation without much practical guidance. And as far as I understood, the goal was to show how stuff should be done in a specific industrial application so at least some problems are “faced and solved for this particular case”, and not to discuss why we do things and what they mean. And this is a completely different thing in my opinion. Perhaps I misunderstood our purpose here, but without a title I would call your work “Lectures on FEA in pipe design” or something similar. And while this has a tremendous value, I’m not sure if it fits the purpose of our work.
Ok, but again I cannot say particular stuff about the nuclear power plant, yet readers who are not familiar with how reactors work will not know why someone would need to compute fatigue of a pipe.
I.e. you mention that each vibration mode has modal mass an excitation reflecting how important it is. But you don’t say where to get those, or how to determine if this is important or not. I would expect you to describe that for the 6 eigenvalues on fig. 5.11 you would explain how to obtain those values and then what is their meaning, and to which point those modes are “still important”. That would have a big practical significance (something I did with imperfections for example, or Nick did in various ways of modeling of his problem).
Point taken. I did not add the values of the mass of each mode not to add pages to the case, but I will need to. They are nice plots actually!
The same goes to choosing stress paths, showing actual linearization and so on… that would be a practical solution. Sure, of only one specific problem, so some commentary of what was not included if things were omitted is important, but only as an appendix and not the main bulk of the work.
Well, this is one of the most important parts of the case study. If you chose bad paths you will get bad results. Again, I will have to stress this.
I think that your work (especially with your style which I like!) would really benefit if you would ditch the discussions about several things (finite volums etc.) and instead describe the model you are solving (one particular case). Showed its BC, loads, what you calculate and how, and then discussion of the outcomes. You are doing this in a sense, but this is all not connected and very vague on details, and I think that your knowledge deserves more than what you wrote.
Sorry, but I am not allowed to do that. I can only comment on general stuff.
I must confess I don’t get your owner and a dog metaphor. Some explanation of it would be nice.
Ok, I will have to rewrite this. Let me know if you get this:
You have two blocks, named A and B. Block A weighs 10 kg and block B weighs 5 kg. If you put the two of them in the same scale, would you swear it will indicate 15 kg? I would not, because if you put two protons and two neutrons together, the result (an alpha particle) weighs less than the sum of the four weights.
What makes us think that weights are always linearly summed?
For someone who critiqued me for the amount of links in my text you used as many as I initially did. So to be fair prepare a webpage with all the resources and post a urt twice as I did, otherwise I will have to admit I did not understand your criticism of what I did at the begining…
You may not believe me but when I raised that point I internally though “hey I might have to order my own links as well” I just did not get down to modify my text until I got enough feedback. Please believe me! :-)
Nodes vs Elements – I completely don’t agree with you on the nodes part. You know in truth those should be DoF right? So we are both wrong here (in some overall sense, I still thing we are both “good enough”). But you also claim you are right and “everyone else” is wrong, while I only use a simplified metric just as you do. It’s obvious you shouldn’t compare element counts of various elements without using computing time as a metric… but I never did that nor postulated it. And various elements may have different amount of DoF per node anyway… so your node method is just as wrong as mine element method… since it’s DoF that are important. They are just a bit ‘abstract” for me so I would rather use simple metrics (like element count or node count) that people can follow (I feel that element count is better here). It’s not that I don’t like that you are using nodes. What I don’t like is your attitude you state in “This is strange because even in college we are taught…”. I don’t know how others look at it, but I try to be as humble as I can (so at times not that much honestly) because God only knows how many times I was wrong before – and when I read your text I got a feeling that author says “I’m good, the rest are idiots”. And I personally don’t like when authors (and people in general) have such an attitude, even though I struggle with this myself since I remember. I guess it’s your own faults that piss you off in others…
Ok, point taken. Those should be DoFs for the solving part, not for the meshing part. And for the computation of stresses it is the square of the DoFs. So it is not even that easy.
I though that I had a sound argument about nodes. Again, I wanted my argument to be written somewhere for future reference, but I can remove this whole part for the time being.
I know your philosophy on black boxes and being transparent and “open source” so I was obviously expected your comments on this. While I don’t consider this to be important or even connected to the practical solution of the problem I also have nothing against that really.
Oh, that is not nice to hear. Again, I failed to make a point.
In general my view is: your writing style is really nice – and it was a good read. Apart maybe from the fact that I expected some high-detail practical knowledge there and instead I found a low-res math approach without any particular problem solved.
You know, there is a part that got pruned to get to the target number of pages about a parametric solution of a tee branch with respect to the diameter of the branched pipe. In that part I link those numerical results to the original theoretical solution of the infinite pipe. In light of your comments aboute being a lecture instead of a case studyy, maybe I should put that part back and prune other stuff. What do you think?
This was a serious let down for sure – especially since you clearly have the knowledge to explain stuff far more “accurately and specifically” in a step-by-step way. So it’s also like a “wasted chance” to help people out.
I chose the subject because
a. I am a Nuclear Engineer b. this case was the hardest one I had to solve regarding mechanical engineering (I had to solve more difficult problems regarding neutronics, thermalhydraulics and control systems)
But this case cannot contain an specific recipe of how to proceed. I wanted to illustrate the general steps that someone willing to undertake a similar case has to take into account, but particular details are out of scope because
a. I cannot share them b. every plant is different
What you wrote definitely has a lot of value, don’t get me wrong. It’s just that as far as I understand this is not a case-study but rather a lecture. I’m not sure how Nafems will treat it (I don’t believe they are super strict to be honest, but still this is borderline in my view). But even if Nafems doesn’t case… I do, and this is why I wrote that.
And I appreciate you having written all this stuff.
This might be a harsh I suspect, but I really hope that I managed to point out where I am coming from in my opinion. By this I mean, I believe in the reasons for which I wrote this, even if the words I used my sound to harsh or impolite (for which I apologize, it was not my intention to offend).
None take. Thanks for all of this.
I hope you won’t hate me for an honest feedback, and of course feel free to disagree with me!
Not that I completely dissagree. In some cases there are only details of difference, in some cases I cannot do it even if I wanted and in some cases I was not able to explain my points. I will try to re-think all of the issues here.
I think that in the end the goal is, that we can produce the best book ever for the “street credit” – and I’m sure we can get there 😊
All the best Ł