Customer Reviews

Programming the Finite Element Method

5 star:		(5)
4 star:		(2)
3 star:		(3)
2 star:		(0)
1 star:		(0)

 

 

This book is an excellent textbook for advanced graduate students as well as the theoretically oriented practitioner with considerable programming experience. However, it should not be considered as a text for a first course in finite elements. Those hoping to learn the fundamentals of the FEM from this text will be disappointed. Other texts, such as Logan's "A First Course in the Finite Element Method," are better suited because of the hand examples along with the step-by-step worked out procedures. Logan's example code is not as powerful as Smith and Griffiths', but is easier to understand for the beginner. Another problem for practitioners is the exclusive use of Fortran 90. Fortran 90 is extremely powerful and may, eventually, save the Fortran language from extinction. However, most practitioners will be almost as unfamiliar with Fortran 90 as they are with C/C++. Additionally, Fortran 90 compilers are relatively expensive compared with C/C++ compi! lers and either will require a steep learning curve. For the student and practitioner with a good background in Finite Element programming and able and willing to invest considerable time and effort, this book will be invaluable. Geomechanic practitioners will find Chapter 6, "Material Nonlinearity," especially useful. Smith and Griffiths have written a very fine book, but not for beginners. The authors should consider writting a true first course in Finite Element programming suitable for the advanced undergraduate or first-year graduate student. If they do so, the two volume series may become the benchmark which defines excellence in FEM textbooks.

This book does a good job showing how to take advantage of modern Fortran90 to program the finite element method with both direct and iterative equation solvers. In fact, this is the only book available to do so. All other finite element books have code which is somewhat outdated in that they use Fortran77 and only cover direct equation solvers, and do not cover iterative solvers.

The book does a good job with showing "element-by-element" iterative solvers developed by researchers in the 1980's. Unfortunately, the book does not cover the more recent iterative methods such as methods which utilize compressed storage formats and domain decomposition methods.

The most disappointing aspect of the book is that the contents claim that parallel computing methods are included. In fact, there is only a very short chapter outlining parallel methods for the element-by-element iterative solution method. Some results are presented of the authors

research but no details are given at all. The authors appear to be novice parallel programming users who are currently learning to write parallel programs for finite elements at the time of the publication. The book would have been much more valuable if parallel programming aspects were included, this would have truly set the book apart from other finite element books. I suspect that the authors will publish a fourth edition of their book after they have learned how to program the finite element method for parallel computers.

This is an excellent piece of work demonstrating the use of finite elements. The fortran 90 programs used give a hands on experience to develop finite elements for various applications. In addition to the detailed explanation of finite elements, this text makes you appreciate fortran 90 to build finite elements for real world problems. The coding of modules used repetitively , which is an important element for learning to write code for finite elements,is emphasized. Despite its emphasis on engineering problems, the concepts can easily be used for other fields.

This book is "one among the rest". It has been written to learn programming F.E. And it just excells at that. After the second edition, in which the use of Fortran77 made the first approach easier but necesarily limited, this one using the extremly powerfull Fortran90 is the best compendium of programs and subroutines explained in detail. I am using it in my curses of graduate engineers and the succes is impresive. The iterative methods for the solvers, are a wonderful bonus. Profs. Smith and Griffiths have done what all off us, teachers, would have liked to do. Anyone studying F.E. must know it.

This book assumes that the reader already knows the
theory behind the finite element method.
The explanations of the code are limited, but with careful
study, the code is powerfull,
concise and portable. This is by far
the best book available in the open literature for
programming the finite element method. The reader with
background in FEM theory, will find this book extremely
useful, and the code very readable.

I hoped to get complete sources for all the examples and programs in the book, but alas...
Just the routines I wanted to use were missing: lancz1 and lancz2 were omitted.
Probably because of copyright issues.
To be complete you need the (commercial) HSL library and it is not included.
The authors should really include a work-around, maybe write their own Lanczos algorithm...
Very annoying, when you finally have found the routines you want to use in your own software, you find out a routine is missing.
The downloadable sources and all the listings in the book suggest completeness... but no...

I must say I agree with the views of `A reader from Seattle', and moreover, I must point out that in such a book one most definitely expects discussion of the optimum way to program the object oriented FEM using Fortran 9x. Instead, the authors only claim that introduction of OO concepts to FEM coding using Fortran 9x is a `matter of taste', which is unlikely to be true, since there are so many publications that emphasize the virtues of such an approach. Considering the fact that Fortran 9x is not fully OO for a number of reasons and thus obviously the C++ versions of FEM codes cannot be directly converted to Fortran 9x, this would have been a most welcome contribution. I have learned that a new edition of the book is due this summer, and I hope that they address this point as well in that edition.

Excellent content, but wish the code fragments were in a different and easier to read font. One more thing is the use of f90. (powerful as it is) I wonder how many languages I have to learn to do any useful programming. (VB for interfaces, C++ for object oriented hoopla, java for portability issues) Looks like life is too short to be learning only languages. Please give us concepts.

If you are interested in programming FEA, THIS is the book you have to buy, without any esitation. Non doubt about it. It will teach you how to code FE solvers (I should have bought it two years ago...), how to set up the element stiffenss matrices and so on.
Actually it is oriented toward FORTRAN90. Well, 90% engineering software are written in FORTRAN, anyway it is a minor problem: one can still easily figure out the algorithm from the FORTRAN code. Once you know the algorithm, coding in ANY language you know is easy...I've been programming FORTRAN 90 FEM codes for two years, and I think this book is a must, even as a desktop reference. SOmetimes the codes presented in this book seem to be a mere upgrade from F77 to F90.

I believe the 3rd Ed has definitely enhance the purpose of this
celebrate Prog the FEM book by Smith & Griffiths. Up to date, besides the books by Owen & Hinton, this is one of the very best book for begineers in FEM in the "sea" of finite element textbook. In fact, the most powerful about this book is the programs that came with it. Buy it and you'll not regret it.