4

I am writing a paper on an engineering topic. We have an experimentally validated simpler model.

Questions:

  1. The current paper extends the previous model for a more complex problem and explores the area numerically. Is it necessary to always include experimental validation for a numerical model, especially since we have already validated it for a simpler case? Is this enough for publication?

  2. Since the primary domain equation, boundary condition and initial condition is the same (what varies between the simpler model and the more complex model is the parameters in the domain equation are more complex), is it necessary to include the boundary condition and initial condition in the paper as well, or can I just refer to the previous paper/include in the appendix? I am trying to reduce the size of the paper, and make it more readable.

  3. The simulation methodology is more complex, because of the increased complexity of the domain equation. Does it help if I include the simulation methodology in the paper? Or should I just put it in the appendix?

4
  • It is not clear from your question 1) what is the focus of the paper - simulation methodology?, 2) what one needs to reproduce your results. Once you decide these, the answers are pretty straightforward. – Greg Aug 13 '16 at 18:01
  • @Greg, Thanks for your comment. Theres a technique in analytical chemistry, lets call it A. Technique A is pretty complex and new. Maybe why nobody has attempted to model it using fundamental physics. I developed a model for a simplified case of technique A, and validated it using experiments using our MEMS device. We have published this. Now I extended this simplified model to the more complex, fully fledged version of Technique A. – alpha_989 Aug 13 '16 at 19:42
  • So this paper covers 2 things 1. It extends the previous model to its fully fledged version 2. Technique A has a lot of design parameters which influence its performance metrics. I did a sensitivity study using the model. I chose the most sensitive parameter and described how and why this parameter influences its performance metrics and how it varies from current understanding. – alpha_989 Aug 13 '16 at 19:42
  • 1. If so I would say the focus of your paper is the simulation method, ergo it would be really strange if you were not be specific about the details. If someone wants to read your paper, he/she will read it exactly for these details. 2. You need to include as much so anyone who wants to reproduce your numbers, could do it without wild guesses what you done. If you think those are too minute details, include them to SI. – Greg Aug 14 '16 at 15:34
5

Is it necessary to always include experimental validation for a numerical model?

Typically, yes, this is necessary. When you are going to publish an altered model (e.g., a more complex one), you need to demonstrate how the modifications play out in an experimental setup.

Is it necessary to include the boundary condition and initial condition in the paper as well, or can I just refer to the previous paper/include in the appendix?

You may choose to either refer to your previous paper, or include them in your new paper (or do both). It is perfectly fine to refer to your previous publication, e.g., when you are trying to reduce the paper length, as long as the conditions are identical. However, when you want your new paper to be self-contained, or your previous paper was at a more obscure venue, it might be wise to include them in the new paper.

Does it help if I include the simulation methodology in the paper? Or should I just put it in the appendix?

It is impossible to decide that without knowing the specifics, the journal's policy, and what is usual within your field.

3
  • Thanks for your comment. I also had a feeling that experimental results would definitely help after reading multiple papers in the MEMS journal I am thinking of submitting to. Unfortunately due to various constraints (financially, time-wise), it will take a while to demonstrate the validity of the results experimentally. Its very expensive to fabricate another set of devices with the characteristics I am proposing. Do you think writing the paper in the form of a letter has a higher chance of publication, instead of a full paper? – alpha_989 Aug 13 '16 at 23:18
  • I understand from reading this thread.. that letters tend to be ok with somewhat incomplete data researchgate.net/post/… – alpha_989 Aug 13 '16 at 23:18
  • You could always try to submit it as a letter, and decide to perform additional experiments when it gets rejected. – Danny Ruijters Aug 14 '16 at 7:34
2

As with any scientific experiment, you should include sufficient detail to allow somebody familiar with the field to reproduce your work. Many papers describing numerical modelling fail to do this, but that is the standard that you should aim for.

For most models validation is vital if any trust is to be placed in it, therefore that should also be described. Whether the validation performed on the simple model is still applicable and adequate for the more complex model is a matter for your (and your reviewers') judgement.

If details of the model or its validation are described in another publication, it is fine to refer to that - duplication of the information is not necessary, although a short summary might be appreciated by readers.

1

Consult the journal you are intending to submit your paper to. Usually they have a set of general guidelines on how they expect you to structure your article. Also look at previously published papers in the same journal.

1
  • Thanks. I did look in the journal, but couldnt find any paper which had a similar story as mine. maybe I didnt look enough, so I will look more. – alpha_989 Aug 13 '16 at 19:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.