<![CDATA[CFD  RevOlutions - My Blog]]>Sun, 22 Nov 2015 02:11:20 -0800Weebly<![CDATA[Pre Processing Resources - Learning how to mesh]]>Thu, 01 Nov 2012 17:29:43 GMThttp://cfdrevolutions.weebly.com/my-blog/free-online-gambit-course-lectures-from-learncax
LearnCAX, which is the academic/education related arm of CCTech has released a lecture  series on Gambit.

They are videos with collated information on power-point slides accompanied by a detailed audio discussion and description.

You will have to register on the LearnCAX website to access the tutorials, and this hardly takes a few minutes and brings the benefit of being notified when LearnCAX release other material/news. LearnCAX is based in Pune, but accessible globally because they offer online courses related to CFD, and CAD (and other things). Check out their website.

Another interesting resource is the 'Turbo-engineer's'  Youtube channel, which has over 37 videos on meshing using commercial software (ICEM, Gambit, etc.) The author has also initiated Linked In groups where meshing problems can be posted. 
Go to the Gambit Course on LearnCAX
Turbo Engineer's Youtube Channel
<![CDATA[Completely Other Things :  Precious Paws Foundation]]>Thu, 01 Nov 2012 16:47:54 GMThttp://cfdrevolutions.weebly.com/my-blog/completely-other-things-precious-paws-foundationPicture
Click to visit the site
While chatting with a schoolmate and friend, I came to know she was working with a non profit organisation called the Precious paws foundation. 

Well, having discovered the power of Weebly, and considering that they already kind of had a lot of content developed over a year - I volunteered to design their website.

Check out the Precious Paws Foundation Website. This site is designed on weebly and their service provider's DNS settings were changed to point to the IP assigned by Weebly. The little graphic you see on top has been designed using good old MS Powerpoint (Sorry, I'm just a beginner !! , it works..). THe picture describes what the organisation does.

I must highly recommend Weebly, it's easy and quick to get up and running quickly. There are of course limitations. But some decent functionality can be added using the HTML sandboxes. The server is quick and I've never seen it down.

 It took me a couple of hours set up the bare bones, using existing content on their FB page. I enjoyed it, and therefore offered to help develop and maintain the website, during my free time. Feel free to visit and mail your comments.

<![CDATA[Karalit and the Immersed Boundary Technique]]>Wed, 01 Aug 2012 19:31:21 GMThttp://cfdrevolutions.weebly.com/my-blog/karalit-and-the-immersed-boundary-technique
I came across an interesting article on the Develop3D website, regarding the Immersed Boundary, meshless technique being used by Karalit in their CFD software. Their FAQ section is worth visiting. Posts about the above are available throughout popular forums like linked in.

They use a novel meshing technique where a structured mesh is super imposed on the existing mesh. The algorithm is claimed to be superior for transient physics and moving meshes. I hope this can be experienced in the 2D version. 

Their 2D CFD software, named CFD vFD 1.2 can be downloaded for free by providing a few details for a registration form. It is a small package, 25 MB in size and they claim no limitations on the number of cores that can be used for the computation. It would definitely be useful for educational and experimental purposes. 

OS compatibility : There are versions for different operating systems, including Windows and Linux versions. Windows 7 and Ubuntu, Open Suse were on the list, with x64 versions being available.

I must say, a wealth of information seems to be available if one types immersed boundary methods on Google. Though a summary can be found on Wikipedia, the original paper by Peskin can also be downloaded, which details the numerical concepts.
----------------------------------------------  - - - - - -  -
Update :
Check out this article on Another Fine mesh , where John Chawner has some updates from ANSYS and CD-Adapco criticizing the Immersed Boundary Method. I agree with John's conclusion - "let the software stand on it's technical merits". The best evaluation of a simulation tool's performance is by comparing it with experimental evidence. 

Quick Links

Immersed Boundary techniques on google
Immersed Boundary method on Wikipedia
<![CDATA[Useful CFD Tips and Tricks on the Innovative CFD website]]>Fri, 27 Jul 2012 20:37:08 GMThttp://cfdrevolutions.weebly.com/my-blog/useful-cfd-tips-and-tricks-on-the-innovative-cfd-website
The Innovative CFD website, has a section with articles themed on tips and tricks as well as the common questions posed by CFD users. It covers a large part of the aspects dealt with with respect to CFD simulations.

Worth checking out, whether a beginner or experienced !

URL  : http://www.innovative-cfd.com/cfd-analysis-tips.html
<![CDATA[ANSYS Resources]]>Fri, 27 Jul 2012 06:47:11 GMThttp://cfdrevolutions.weebly.com/my-blog/ansys-resources
Having recently joined the corporate environment, it was pleasant to see first hand, the kind of support that is available to customers via the ANSYS customer portal; tutorials and help files are readily accessible. Though a lot of the material is available online via other sources, the latest updates can be found via the portal.

ANSYS products are a commercial (and costly!) software suite and access to a machine with a working installation is required. I will try to focus on resources which are openly accessible and useful in beating the learning curve. This should minimize your search for good tutorials and provide a jump start; An ANSYS section has also been added in Resources, collating the individual resources mentioned on this blog. 

It is common in Universities to host tutorial files within their intranet and have the help files set up locally, rather than provide access to the Customer portal. The academic versions of ANSYS provide limited modules. There are other Universities and researchers who partner with ANSYS in releasing books and official tutorials that cover the ANSYS suite of products. It's important to know as much as possible about the tool that is being used to perform the simulation and generate the results on which we base our analysis.

The tutorials put up by Cornell University are time-less in the way they have usually been the starting point in academics. The above link has sections on structural analysis, CFD ( using Gambit and Fluent) as well as MATLAB. A great set of tutorials with categories starting from the beginner to the advanced level is accessible at the website maintained by the University of Alberta, which also has a post processing tutorials section, though the sections are focused on structural analysis.

Gambit is now a dated pre processing application; it may be more useful to get familiar with a commercial tool like ICEM  or even open source tools like Salome, Gmsh, Engrid. Pheonix Analysis & Design technologies have a periodic electronic publication called The Focus, which is available for free and all the past articles are also archived. This magazine has articles that are very useful and unlike the usual kind of articles that are found on the web, they are professional and meant for the general ANSYS user, with interesting aspects explored.

ANSYS.net is a highly reccomended source, both by The Focus magazine mentioned above and other websites as well. It also contains free to download macro's and tips and tricks regarding ANSYS. However, there doesn't seem to be much of CFD related applications in it. Writing macro's is also an important skill to cultivate, as it aids in automation and saving time.

One should pay a visit to the official ANSYS website as well as blog, which is more often focused on events where the ANSYS products have been used along with the occassional foray into emerging areas where ANSYS is being used. There is also a tips and tricks section which is useful. The free webinars hosted by ANSYS, on various topics are posted on this blog and it includes a tips and tricks section too.

Xansys is a ANSYS community portal that is fairly active. The forum seems more focussed on FEM related structural applications, which is what ANSYS was known for initially. The most useful feature and the reason why Xansys is being mentioned is the ANSYS Links page. A few of the links are outdated and don't work and the main links are posted in this article. This page contains links to the most prominent and prolific resources related to ANSYS, covering reviews, tutorials, product updates and information about other software applications and the industry as well. 

There is an active ANSYS forum on CFD-online as well, with subforums categorised based on products, like CFX and Fluent as well as another related to the Pre processing.

Additional Links :
(This section will be updated as and when I find new resources )

Quick Links

<![CDATA[Open Source Software]]>Tue, 10 Jul 2012 15:04:02 GMThttp://cfdrevolutions.weebly.com/my-blog/using-open-source-software
The notion that Open Source Software (OSS) is difficult to master or apply is slowly melting away. Infact, current 'mainstream' Linux distributions are developing along the lines of being very user friendly to the extent that the interface affects the core functionality and flavor of the distribution. The latest Ubuntu Unity version for example has impressive graphics and several bells and whistles attached to the UI to provide a Mac like feel, using Compiz. The above is in comparison with Ubuntu 10.04 on which CAE-Linux is based. It may interest you to know that the Mac is a customised Unix based OS. 

Which Linux Distribution ?
Linux distributions basically have different flavors or interfaces. An analog would be : the different versions of Windows. However, I believe Microsoft intend to do away with the versions in Windows 8. The most popular interfaces on Linux are the Gnome and KDE. Ubuntu and Kubuntu are examples of the Gnome and KDE flavors. SUSE is an OS further customised available with a KDE base or Gnome. I think SUSE has slightly tougher learning curve than Ubuntu, though the workspace customisation options are simply delightful. Ubuntu is known to make operating systems extremely user friendly. For example, Ubuntu 10.04 LTS is a lot like a tweaked Windows XP. It's something that is easy to get used to. 

There are Linux distributions that are tailor made for specific purposes. The website Distrowatch covers the latest releases, news and reviews related to Linux distributions and worthy of a visit. There are free Linux guide books available for download on the website, but there is a lengthy registration process.

Anyhow, there are enough  easily accessible resources on the internet to master Linux. Have a look at the Learning Linux Section as well.

Personally, I have also found SUSE's enhanced KDE interface very interesting, with the concept of grouping folders via panels. SUSE however, takes a noticably longer time to startup and definitely feels slower than CAE Linux. However, on downloading the KDE desktop and interface via CAE Linux, there was a problem of missing menu items. In this case, it was the all important CAE Linux Menu. The short cuts were available, but had to be individually typed in the search bar, and the solution was to add a panel with the short cuts on the desktop.

Recent articles have alluded to the thought process and possibly popular opinion that Microsoft has been losing ground to the Apple products' sheer might and appeal, rather than Open Source or Linux. The ultimate appeal of OSS is that it is free and gives one the possibility of actually controlling applications, their output and automate any aspect at all.

Conventionally, the industry was limited to writing User Defined Functions (UDF's) and hooking them up to commercial solvers. It should also be noted that the majority of servers and supercomputers were and even now, usually run on Linux. The most successful case of Linux being used commercially is that of Red Hat Enterprise Linux. They provide the source code but not do not provide the compiled binaries. Red Hat changed the status quo by providing Support (and relatively expensive support), the much needed ingredient to bolster OSS. CentOS is a free distribution based on the Red Hat source code. It is known to work best as a Server based OS.

As a conclusion to the kind of OS you should choose : Ubuntu's distributions have the most user friendly interface and is ideal for a beginner. CAE Linux is one distribution to consider in that case. SUSE and Geeko CFD is also another option, but is known to make the installation of packages, a little more complicated than the well known apt-get command in Debian and Ubuntu or the very friendly package manager. Once you have gained a footing, Fedora/Debian/SUSE may be the distributions to turn to for more power and maximum stability. It really depends on the type of environment you deploy the system and whether it's a desktop or a Server.

'The best software' ?
Let us say, one does take the plunge OSS, to understand the mechanics and the codes solving all our simulations. Often, irrespective of OSS or Commercial applications, when it comes down to choosing the software package, questions and possibly involuntary thoughts occur related to 'choosing the best software'. Such questions can especially be seen in forums regarding 'which CFD software is the best'. This question does not have any direct answer. In CFD, to the largest extent possible,the code's performance should strictly be measured against valid experimental results. This should hold the same of any code. It's fair to say that OSS is not 'better' than Commercial software. The vice versa also holds true. These are just different platforms with their characteristic features. Contributions and opinions of experienced members in forums indicate the above as well. Infact, the performance of a solver, solving a particular kind of physics must be evaluated rather the software package itself. For example, Gambit, despite it's limitations was very widely used and is still being used by both academic and industrial users. The majority swear by commercial pre processors like ICEM, ICEPAK, HyperMesh etc. These tools are known to get the job done, which  is very important in the industry. It would seem that User Friendliness, is an aspect that comes in at the very end. Most software that come out nowadays, are however user friendly enough. 

The question of OSS being more difficult is more often because of possibly steeper learning curve, which in turn is rather an issue of unlearning the things we're used to. Somebody familiar with computers in an intimate manner could get used to the Command Line interface and Linux pretty quickly. 

Discussion :
For example, I recently had a go at cross checking an OpenFoam (OF) installation on the latest Ubuntu Unity version, while providing an overview of the key elements that make up an OF simulation. Such a check would simply involve running blockMesh and the solver, say icoFoam and then starting Paraview to post process. 

I found that the new shortcut bar on the desktop was a little bothersome to get used to, but it was definitely not the kind of OS anybody need to be afraid of. However one should remember, running OF and installing different applications will very often require libraries and an active internet connection is the most convenient way to get up and running quickly and relatively painlessly.

This is where the advantage of a pre-compiled distribution like CAE-Linux or Geeko-CFD is undeniable. There are usually no missing libraries and application shortcuts are available in the CAE-Linux group in the Start Menu. This is quite useful for those that are too apprehensive of the Command Line Interface.

An effort is being made to write up this experience, information and relevant resources on CFD-Wiki on CFD-Online, the most popular portal of CFD. The section under development is titled 'The Open Source Section'. The more personal information and views will be put up on this blog. 

When approaching Linux and moving away from GUI cushioned applications with heavy commercial support, one should remember that even seasoned Command Line users copy & paste commands and look up commands as well, frequently. It's more important to read up on things and take a plunge by getting an Open Source distribution at your disposal. Preferably something like CAE-Linux. I've experienced no issues regarding missing libraries so far.

Of course, eventually, one can anyway come up with a more suitable distribution if and when required - by learning how! Ubuntu 10.04 has inbuilt options to customise a distribution. Another point to note is that using a Linux Operating system with a Command Line interface and considering the overall architecture/methodology on which OF is built on, requires a thinking pattern/approach that is more proactive and focussed on more intelligent thinking and maneuverability.

However, there are studies, not more than 3 years old that evaluate the preprocessing capabilities of applications like Salome, GMsh and Elmer. The author of one such paper, deemed that Salome is the closest that these OSS come up, against industrial veterans like Ansys and Altair. Results have been obtained from OF which are reliable, but time is of essence , in terms of being suitable to industrial applications. Few have the expertise and the time to devote to implementing a system and also taking care of debugging and maintenance.

Conclusion (.. for now )
In terms of uncomplicated everyday use, like emails, web browsing and media, OSS and operating systems can definitely serve your purpose well. The problem arises when you have to extensively use MS Office applications. I intend to find a work-around for this bottleneck soon. Antiother issue is support. Not many are fully conversant with Linux OS's and one has to be prepared to exercise his brains, looking for solutions on the internet.

In terms of CFD and CAE, open source applications definitely offer performance that is noteworthy of comparison with commercial solvers. But the Pre and Post processing capabilities, which is extremely important, still have much to cover. The biggest strength is how things can be automated using simple BASH or Python scripts.

Quick Links:

<![CDATA[Is CFD an Exact Science?]]>Thu, 07 Jun 2012 10:08:08 GMThttp://cfdrevolutions.weebly.com/my-blog/is-cfd-an-exact-science
"CFD can be exact but CFD does not have to be exact to drive product development decisions"

This is a great article, from Informative Design Partners, that is also the topic of a linked in discussion. The article addresses a question that is probably ancient. But with the advent of CFD, I think it's importance is certainly strengthening, and a lot more businesses are concerned about the answer. The URL's to the blog article and the Linked In discussion are in the adjacent column.

I believe that a thorough knowledge of the limitations of a technique and assumptions made in formulating it, are critical factors contributing to the accuracy/ reliability of a CFD analysis. That's also an area where experience comes in. I should probably say, a lot of experiences. 

I've faced such a situation myself, with an overwhelming amount of simulation data : case studies with very minute observable differences; where the question arises, how much of the data should be plodded through.  Especially considering the stranglehold placed on the Time you can dedicate to achieving 'perfection', the ultimate objective of exact scientific application. It takes a certain amount of common sense, and intuition to pick and choose the data that makes the biggest impact relevant to the focus of your study. Intuition/Hunches are not infallible. That's where common sense and a scientific perspective of thorough but selective evaluation comes into the picture. 

Does CFD then fall into the category of an exact science ?
Being a CFD man myself, I'm more inclined to give in to my bias and label it's current status as the art of Refined Engineering Application. But make no mistake, the ultimate objective of CFD is to become an Exact Science. I believe several relatively simple case studies do exist which quantify the assumptions made while performing a CFD analysis. More on this later.

See the Popular CFD Blogs and industry news section for the RSS feeds of popular blogs including leading software companies. 
<![CDATA[Pump Design Source]]>Wed, 06 Jun 2012 20:09:27 GMThttp://cfdrevolutions.weebly.com/my-blog/pump-design-sourceThis is a website that offers comprehensive information including videos, applets, an extensive Glossary as well as tips on pump design. The entire material is also collated and written down in the form of a free-to-download ebook. The notes are not very involved from a mathematical aspect and has turned out to be a pleasant and interesting read. 

You may get a warning about malware from this website. However, I've been exploring the website and nothing seems to have happened yet. This is definitely, a great source of information on pumps and the physics behind it's functioning as well as design. The site tries to provide you complete background theory as well to enable you to understand pump design and also provides links to good material in other sites, including those of manufacturers. 

Check out > http://www.pumpfundamentals.com 

See the section All About Pumps : A collection of web sources related to Pumps, including fundamental concepts, Industry news and very interesting technical magazines and newsletters that offer free digital subscriptions.]]>
<![CDATA[How Grid Shape and Quality affect the solution]]>Sun, 06 May 2012 07:26:05 GMThttp://cfdrevolutions.weebly.com/my-blog/how-grid-shape-and-quality-affect-the-solutionLinked In Discussion - 
URL : How does the Grid shape and Quality affect the solution ?

This discussion has some excellent comments related to the approach of setting up a grid as well as the logic behind such decisions. It also highlights the characteristics and concepts of using different cell types as well as the difficulties encountered in setting up a structured or unstructured grid.

<![CDATA[Verification and Validation of CFD results]]>Sun, 22 Apr 2012 05:24:49 GMThttp://cfdrevolutions.weebly.com/my-blog/post-title-click-and-type-to-editLinked In Discussion -
URL : How many of you do Verification and Validation for your domain? or simply trust what you did is correct? 

Any student of CFD can benefits from discussions in forums, especially when the replies are from experienced and seasoned multi-national professionals practicing CFD. One can always miss certain aspects or perspectives during intense research, or simply may not have come across the issue being put across in a such a succinct manner. Words and the way they are combined and finally presented, have a mysterious way of opening up one’s mind.

I feel it is important to read & participate in discussions that enhance everybody’s understanding of a subject, especially one as complicated (and beneficial) as CFD. Depending on the forum, one ends up exchanging views with an international community, which enhances the reliability of the knowledge gained from these discussions. Check out these Research Networks and Forums

Note: Linked-in Membership is not required to view this discussion, since it was posted in an ‘open' CFD group’.

Summary of the discussion + extracts from notable comments so far :
  • Verification and Validation of solutions are different aspects and both are very important.
    • Verification : Enables one to know how well the solver is solving the set of differential equations, with comparison to expected and analytical results/characteristics.
    • Validation : Solutions are compared with Real Life observations and Experimental results.
  • Numerically correct does not necessarily mean Physically Correct.
  • However, at present, Verification and Validation are not always performed or emphasised on. It seems dependent on the situation and importance of the simulation results i.e what kind of decision is being made based on the simulation and what the consequences of mistakes are (eg Danger to Life).
  • Optimisation involves different tactics from conventional approaches. Here, the focus is on comparing cases which are similar in terms of the boundary conditions and settings and thus shortcomings. i.e one would develop and finalise a ‘Base’ case which has all the refinements and then develop other case studies with exactly the same settings, so as to form a valid comparison and optimisation.