this is the first article of a series that will focus on proper application of the finite element method for solving engineering design problems. the author of these articles (introduced at the end of the article) is an experienced user of advanced fe tools whose insight may be of value to our members.

i. introductions

although fea is based on a very rigorous mathematical foundation, its application is as much an art as it is science. so i hope by sharing my experience in this column, i can be of help to other users of fea technology. fea, like other engineering tools requires careful and continuos practice. in the hands of a novice this tool could prove disastrous.

casual users may not get enough practice to become proficient and economically viable in utilizing this tool. attempts by software companies to automate fea so design engineers with little or no fea experience can use them have so far proven not very successful. some vendors do offer a simplified version of their fea code for first cut trend analysis by design engineers. however, in this series of articles, i refer to standard fea software such as ansys, etc.

what are the prerequisites for a good analyst? a good fe analyst, is a good engineer first, i.e., has a good understanding of the fundamentals of the engineering discipline, can simplify most problems so that they can be solved using closed form equations. an understanding of fea theory is helpful, but not a must.

knowing the fundamentals will be of great help in the beginning when decisions in modeling and results interpretation can not be made based on experience. the help features and the general jargon of fea software also require some knowledge of the basics. there are many textbooks on fundamentals of fea, some targeted at undergraduate level.

a good analyst should also be able to interrogate the analysis output and determine if the results obtained make engineering sense. this is the first line of defense against modeling, input, or solution errors. just because the results were obtained using an expensive fea code does not mean that the problem has been solved correctly. since fea can solve very complex problems, the idiom "garbage in, garbage out " is very relevant here. hence, the engineer needs to bracket the problem by closed form solutions for a quality check on fea results. additionally, one should be able to determine the sensitivity of the results to design parameter changes. this understanding is a prerequisite to design modifications, since without it redesign becomes a costly trial and error exercise.

ii. understanding the physics of the problem

fea, like all tools for engineering problem solving requires a correct definition of the problem statement at first. a correct problem statement requires a thorough understanding of the underlying physical phenomenon involved in the problem. if the physics of the problem is not well understood, or information is incomplete or unavailable, then no tool, including fea can help solve the problem. in particular, field problems sometimes contain irrelevant or erroneous information and any attempt to perform analysis without a thorough understanding of the physics involved will yield irrelevant results.

sometimes the physics of the problem is well understood, but the input data such as material properties or boundary conditions are not fully defined or readily available, forcing us to choose between abandoning the project or developing the missing data through other means. in other instances we face problems that may involve complex and poorly understood phenomenon that cannot easily be solved by any method including fea.

we should also be aware of the temptation to use fea indiscriminately to solve every problem we encounter. new users are more prone to this type of fea misuse. although fea codes have become extremely efficient and user friendly, performing fea still requires modeling and solution effort, plus computing resources. so if the level of accuracy required in solving a problem lends itself to hand calculations, fea is an over kill and sound engineering practice dictates that we use hand calculations.

a correct problem statement also helps us clearly define the scope of the analysis. poorly defined problems and false or unrealistic expectations from fea can backfire leaving an unpleasant legacy that is hard to overcome in an organization. during my visits to client offices i have seen copies of relatively inexpensive fea codes collecting dust.

in a typical scenario an engineering manager reads an advertisement for fea code x promising everything under the sun, from ease of use to unrivaled robustness, all at a modest price tag. a recent college graduate is assigned to attend a few days of training and is soon given the task of supporting production-related issues using this code.

once expectations are not met due to user inexperience, organization's impatience etc., the initial enthusiasm turns to indifference and the software is shelved.

hence, we should not only fully understand the physics of the problem at hand, but be able to discriminate between three categories of problems; problems that can be solved through simple hand calculations, problems that can be handled by fea, and those which could not be solved using fea.

some engineering problems are multi-disciplinary in nature and require well thought out strategies before any analysis is attempted. in majority of problems however one physical phenomenon dominates and the problem can be categorized as structural, heat transfer, fluids, etc. for cases where weak coupling exists between these physical phenomenon, uncoupling them may be appropriate.

for example in a problem involving thermal stresses, effect of resulting stresses on temperature distribution can usually be ignored. so the two concepts of stress and heat transfer can be decoupled in this case. on the other hand in thermofluids problems where fluid and thermal properties are tightly interrelated or coupled, the above strategy would not be appropriate.

some heat transfer problems may involve radiation heating or cooling. radiation is a function of surface temperature and view factors. most fea codes require some manual work for view factor calculations and since radiation and surface temperature are coupled, this type of problem requires an iterative solving procedure which is computer time intensive. therefore radiation should only be included if it is a significant contributor to the physics of the problem.

the engineer still needs to further refine his understanding of the physics of the problem by discerning whether or not the behavior under investigation is time dependent. physics of a steady-state problem is much simpler to understand and simulate, than that of time dependent transient problems. also just because the physics imply the presence of transient effects, one should not necessarily conclude that the problem a transient one. sound engineering judgment can determine if the transient problem can justifiably be reduced to a steady-state equivalent case.

other time dependent phenomenon relates to dynamic effects in structures. effects of inertia, dynamic loads or impact may or may not play a significant role in the physics of the problem. inertia of an accelerating body becomes important in impact. if the mass however is relatively small, inertia can be ignored. in dynamic loading if the time of load application is less than half the fundamental natural period of the structure, the loading is definitely impact.

on the other hand for problems where the time for loading is larger than three times the fundamental natural frequency, the dynamic effects could safely be ignored. however the main difficulty in most impact problems is determining the time of loading.

since impact loads are coupled to the behavior of the loaded structure, we need to examine this interaction to fully define the physics of the problem. localized effects such as friction, plasticity, etc. are not well understood for most problems, so other simplifying assumptions may replace the physics of the problem.

about the author

dr. moazed is an engineering consultant with expertise in finite element analysis. for the past twenty years, he has utilized fea to solve a variety of engineering design problems nationwide. dr. moazed specializes in analysis of rotating equipment such as fans, compressors, pumps etc. he received his ph.d. from the university of tennessee in engineering mechanics and is a registered professional engineer in the state of massachusetts.