Sei sulla pagina 1di 12

Past, Present & Future of FEM

2 Past, Present & Future of FEM

2.1 History of Finite Element Method

1906 - Civil Engineering applications, Structures analyzed by 1-d beams (hand calculations).

1909 – Ritz – Variational Method

1915 – Galerkin – Weighted residuals

1940’s - Courant, Prager & Synge - Mathematical foundation for present form of FEA

1950’s - Argyris, Kelsey, Turner – Direct continuum elements, Aerospace industry engineers formulated
stiffness equations. 100 dofs problem was supposed to be very large at that time.

1960 - Clough coined the name Finite Element method.

Late 60’s - Mechanical industry recognized FEM as useful tool for solving real life problems. Many of
the presently popular FEA softwares were launched in the market.

1980’s - Graphical & computational development.

1990’s – Emergence of low cost, powerful PC work stations & FEA adopted by mid & small scale
industries

Year 1961 1966 1971 1976 1986 2006


Papers 10 134 844 7,000 20,000 infinite!

2.2 Present

Status of FEM as on Today

Theoretical
Practical FEM as its
(At universities & for software Software based FEM
used in Industry
code writing)
[F] = [K] [δ] Preprocessing Solution New Design
Force stiffness displacement Postprocessing - Solid
modeling Cost cutting
[Q] = [K] [T]
heat conduction temperature - Meshing Failure / Optimization
Analysis
[I]= [1/R] [V] - Boundary conditions
Current resistance Voltage

Language used for most commercial software code generation are - Fotran and c++ .

Theoretical Finite Element Analysis

Till 1996 -In India FEM was offered as post graduate course only at few universities

18
Practical Finite Element Analysis

1997 and onwards- introduced at graduation level

Theory courses deals mainly with


1) Various methods to derive [K]
2) Assembly of [K]
3) Solution techniques

Element Stiffness Matrix

Stiffness matrix is like password or PIN to the treasury of FEA. [K] stiffness matrix , basic property of
element depends both on geometry & material.

If you refer to any theoretical text book on FEM, most of the volume is devoted for methods describing
how to determine [K] or stiffness matrix of the element & its assembly. There are 3 methods for
deriving [K]

1) Direct Method - Easy to understand but difficult to program. Its not used for commercial software
code generation.

2) Variational Method - Rayleigh-Ritz Method : difficult to understand moderate from code writing
point of view.

3) Weighted Residual Method - Galerkin Method : difficult to understand but easy from programming
point of view. This method is used in most of the commercial softwares.

What is stiffness ?

Stiffness K = F (Froce) / D (Displacement)

Stiffness is defined as force required to produce unit displacement. It depends on Geometry as well
as Material properties.
Steel Aluminium

F F

Consider 2 rods identical geometry but different material say Steel & Aluminum. Which one is more
stiffer ? Steel, because of its higher elastic modulus.

19
Past, Present & Future of FEM

Steel Steel

Now consider material is same i.e. steel but area of c/s is different for two rods. For this case the rod
having more area of c/s is more stiffer.

Tesnsile stiffness = Ktension = AE/L


Bending stiffness for cantilever beam = Kbending = 3 EI / L3
Tortional stiffness = Ktorsion = G J / L

Is a Pentagon or hexagonal or any arbitrary shape element possible?

Why in FEM we always see only limited regular shapes like quads, tris & brick, wedge & tetra only for
years & years.

Finite Element started with Triangular element. Later Quadrilateral element was formulated.
Theoretically one can have any shape for the element. Only thing required is to know the stiffness
matrix. But in practice since we are able to mesh any kind of geometry right from structural
components to Aerospace with available 5 or 6 basic shapes, need for other shapes is not really
felt. If it is so then why quad element was introduced at all, was it not possible to mesh with only
triangular shape.

-The need for quad element was felt from result accuracy point of view. Triangular element being
stiffer results in less stress & displacement if used in critical areas or independently for entire geometry.
Generally at professional level a Tria – Quad combine mesh is used, restricting % of tria elements to
5.

Why number of tri elements are restricted to 5 % & avoided in critical areas?

Vs

• Quad element is more accurate than triangular element.

20
Practical Finite Element Analysis

• Tri element is stiffer than quad, results in lesser stress & displacement if used in critical locations.

• Consider any standard problem of plate vibration for which analytical answer of natural frequency
is known. Create two fe models based on only tri elements & only quad elements (same element
length). Tri mesh will show higher value of natural frequency (vn = √ k/m) & will deviate more
from the exact answer.

2.3 Software based FEM

For using any commercial software there are 3 steps namely –

1) Preprocessing – Consumes most of the time out of the three.

2) Processing (or solution) – just click on “Solve” & it’s the software trun to do the job.

3) Post processing – Result viewing & interpretation

Step 1 - Pre processing

a) CAD data

b) Meshing (or discretization to convert infinite dofs to finite one)

c) Boundary conditions

When FEM & commercial software codes were in early stage CAD, Meshing & analysis all used to
be carried out by a single engineer only. Soon it was realized that separation of the jobs & forming
dedicated groups i.e. CAD group, Meshing group & Analysis or calculation group is necessary for
optimum output & efficiency.

CAD & Meshing - There are specialized softwares for CAD, Meshing & Analysis. CAD& meshing
consumes most of the time. For example - Typical time for a single person to model engine block
4 cylinders is around 6 weeks & for brick meshing again 6 weeks (For tetra mesh its around 2 to 3
weeks).

Boundary conditions - consumes least time but it’s the most important step (typically considering
10 load cases its about 1 day job). 3 months hard work in meshing & CAD data preparation of engine
block will be wasted if boundary conditions are not applied properly.

After completion of preprocessing i.e. above three steps CAD, Meshing and Boundary conditions,
software internally forms equations like [F] = [K] [δ].

Step 2 - Processing or Solution

In Preprocessing hard work from user is required while in processing its turn of computer to do
the job & user has to just click on solve icon & roam around ! Internally software carries out matrix
formations, inversion, multiplication & solution for unknown like displacement & then find strain &
stress.

To we are using FEA just because of availability of computers. FEM has been known to mathematicians

21
Past, Present & Future of FEM

& engineers right from late 50’s but since solving so many equations is just not possible manually
hence in true sense FEA got reorganization only after emergence of computers.

Step 3 - Post processing

Post processing is viewing results, verification of results, conclusions & thinking about what steps
could be taken to improve it.

Just consider a simple example which involves all the above steps

Probably at the moment you sitting on a chair or stool & reading this book. In this sample example
the stool it self is analyzed for stress & displacement for a load of 200 kg (assuming it could be used
for sitting as well as supporting any object up to max. 200 kg wt.)

a. CAD model

b. Meshing and Boundary Conditions

Load = 200 kg Element type - Tetra 10


Nodes = 60,176
Elements = 34,637

Uy = fix, additional min. dofs fixed to


avoid rigid motion

22
Practical Finite Element Analysis

C. Solution & Postprocessing

Type of analysis - Linear Static, Material - Steel

No. of equations / unknowns = no. of nodes * dof per node = 60176 * 3 = 1,80,528

Max. stress (5.97 N/mm2) >> yield stress (250).

Analysis conclusion

• The stool is over designed.


• There is scope for cost reduction by reducing thickness of legs, top plate etc. (assuming there is
constraint that material has to be steel only)
• Material could be changed to low cost alternative (having yield stress > 6 N/mm2) {assuming
constraint is design dimensions should remain as it is}
• Max. stress at the sharp corners which could be reduced by providing smooth fillet or addition
of stiffeners in the localized region.

2.4 Practical applications of FEA in the Industry

There is a dedicated team of engineers responsible for CAE. They receive following types of job
orders

• New design
• Optimization or Cost cutting projects
• Failure analysis

New Design:

New concept or innovative kind of designs for which design engineer has no prior experience &
is not sure about how the component will perform in the field. In automobile industry, when new
version of existing vehicle is launched (upgraded version), most of the components are quite similar
to the existing one (scaled proportionately) & innovative kind of components are usually not more
23
Past, Present & Future of FEM

than 15 %.

At least for initial run of this category of jobs are easy for CAE engineer i.e. sit with design & experienced
test engineer to decide boundary conditions & run the analysis. Real challenge starts only when the
prototype is prepared and test & FEA result correlation process starts. After the correlation various
permutations & combinations could be carried out to make the product better and optimum from
cost as well design point of view.

Cost cutting or optimization project:

At the moment Indian (& Asian) Auto sector is experiencing boom - boom but from 1997 to 2003
there was a slack. During the period most of industries were busy with cost cutting measures for their
survival.

In Indian market till late 80’s, same kind of vehicles were running on the road with out any change
for decades! (do you remember old designs of ambassadors, fiat, jeeps & heavy bulky & noisy trucks).
These designs were transferred to Indian companies in 50’s & 60’s from their overseas collaborators
(mainly American & European’s). Design philosophy was different that time i.e. design for infinite
life. But slack in the market and emergence of new tools like CAD/CAM/CAE, new cost efficient
manufacturing techniques, and availability of new low cost materials forced auto manufacturers to
adapt to the changing circumstances via optimization of the design.

Suppose selling price of the product is Rs. 100 & actual manufacturing cost is say Rs. 60. Reduction of
cost even by say Rs. 1 by using CAD / CAM / CAE (reduction in thickness, change in material etc) will
add to profit of the company.

Earlier days design philosophy used to be – Design for infinite life


• Survival for years
• Heavy & oversized components
• High cost
Now a days design philosophy is - Design for warranty life
• Use & throw concept
• Life > warranty offered by the company
• Light weight component
• Low cost
• Additional source of income for the company after the
warranty period (i.e. after sales services)

2.5 Failure Analysis

Warranty : Every company give warranty on its product. Company is under legal binding to replace
the component failing within warranty period free of cost. Its not only additional cost which is
incurred on the company but also the bad name in the market.

Probable reasons of failure

• Inproper process
24
Practical Finite Element Analysis

• Manufacturing defects

• Faulty material

• Environmental conditions

• Weather

• Road condition

• Design abuse

• Design problem

Warranty yard meetings & throwing the ball in others court :

Every medium or large scale industry has a warranty yard (a separate place where all the failed
components record is maintained systematically and failed components are displayed). Once in
a month managers from different departments meet at warranty yard and discuss about failed
components and corrective actions. Whenever something becomes critical (in industry something
becomes critical & things starts moving only when its being highlighted by VP or Director or President
or CEO of the company), say current meeting is to discuss large number of warranty claims for some
xyz component. In Industrial world, it’s an open secrete, nobody is interested in taking responsibility
for failures and finding the remedy on their own. Instead we play the game as per the rule : “Always
be safe throw the ball in other departments court”. In atleast warranty yard meetings R&D or Design
departments “the soft targets” are not that lucky in ball throwing games and most of the times hold
responsible for all the failures. In side the company its Design and R&D & outside the company its
service engineers who are the soft targets & takes most of the blame

Please join a hypothetical meeting of different departments at warranty yard as an independent


observer. Scene and atmosphere is tense due to “critical” nature of the problem and here we start
interesting meeting.

Manager, Manufacturing - We are strictly following all the instructions as it is mentioned on the
drawing released by design / R&D, I am confident its not at all a manufacturing problem, R&D is the
clearly responsible for the failure.

Manager, Purchase – We are working so hard you know! Moving around the globe to find the best
quality of material / vendors that too at lowest possible cost as per the requirement by design /
R&D.

Manager, Quality - I guarantee you that all the tolerances, surface finishes, paint quality etc are as
per standard.

Manager, Marketing /Sales - Marketing people are born showman and Bol Bacchans (expert giving
speeches). Automaticlly leadership of the meeting goes to them. They fire Design / R&D manager
claiming that its because of their poor work, marketing and sales team has to adopt to face saving
tactics and manage the things with customers diplomatically in the market. Our company is investing
millions of dollars on Design & R&D, people are given such a handsome salary, they are always seating
(what they actually mean is sleeping!) in AC’s in front of high capacity computers but see what is the
output?
25
Past, Present & Future of FEM

Managers - Process, Industrial engineering, Maintainance - Other dept. people feeling naturally
happy because the finger is not pointed towards them, they with a pleasure join others to fire left &
right poor R&D and design guys.

CEO / MD / President - Now comes the final blow from the CEO or President of the company who is
less interested in knowing actual technical details or root cause of the problem and how much time
is required to solve a given problem: orders R&D chief “See, I want solution to this problem within
1 month, you do whatever you want but I want to hear that the problem has been fixed by next
meeting O.K !”

R&D /Design Dept. - R&D head calls a meeting of his subordinate managers & order (& transfer
them to all the tension, blame & abuses presented to him in the meeting) that the work should be
completed within 3 weeks {thinking that if 1 month time is given to them, then they will take more
time & by doing it early he can also prove his efficiency (performance appraisal) to the CEO !}. After
the meeting R&D head feels relaxed & relieved since he has done his job effectively. Usually this
type of work is carried out by a CAE engineers. In the hierarchy there are 2 to 3 superiors & when
actually the job comes to him every superior using his own factor of efficiency gets about 4 days or 1
weeks time limit!! Now all the tension is transferred to CAE engineer. The boss and managers would
do the constant follow up every now and then and irritate the engineer like a credit card or banks
telemarketing agents!!

Now let us continue the story with some of the (hypothetical) case studies

Truck Steering knuckle failure analysis

In one of the organization steering knuckles were failing in large numbers at the location as shown.

Crack

Small step due to manufacturing


error

Steering knuckle fig.

CAE engineer was not aware about the warranty claim, he has been just communicated that he has
to analyse steering knuckle. R&D, CAE facility was away from manufacturing location & all the details
could not get communicated to the responsible engineer properly.

Start the job : CAE engineer got the 2-d drawing from data room & prepared CAD model, carried
out meshing and analysis and then submitted an impressive colorful plots report. The failure location
as per CAE report was away from the field failure location. From the analysis CAE engineer concluded
26
Practical Finite Element Analysis

that since stress level is below yield limit, it is safe.

R&D submitted the CAE report, after going through the report, in the meeting other department
heads just fired R&D manager saying that real life failure is at different location & your team is saying
its safe, no need of modification.

Later the CAE engineer was informed about all the proceeding & gravity of the problem. After going
through the failure reports & observing failed components, some of the CAE team members visited
manufacturing site and machine shop. Luckily one of the member detected very small sudden step
at the location of failure on the some of the components due to faulty tool settings on one of the
turning machine. There were 8 turning machines & setting was faulty on only one of them. CAD
model and meshing was modified so as to accommodate this step. Analysis showed exactly same
location of failure as per the field report i.e. at sudden step. Clearly this was not a design problem but
problem due to defective m/c setting.

Starting point or input for CAE is CAD data as per the drawing & not how the actual part is
but for failure analysis checking actual part geometry, visiting the shop floor, understanding
manufacturing process, going through the failure reports & observing failed components
should also be included as basic pre-requisite for accepting analysis job.

Climatic conditions:

A road construction equipments & farm equipments accessories manufacturing company was facing
failure problems. Region wise segregation showed failures mainly from specific states i.e. mainly South
India only. Same design, same components working successfully in rest of the India. This is because
South Indian soil is acidic and causes faster rusting (components are merged in mud & water most of
the time). Rusting eats/consumes material & reduces thickness and life of the component.

Should a CAE engineer be assigned the responsibility of rusting analysis. Could FEA help in solve
rusting problem. Answer is No! Instead a material science or chemical engineer or paint shop expert
should be consulted. Since this problem is a specific to a particular region. One of the solution could
be to separate out batch of vehicles meant for South India and provide special anti corrosion primer
(coating) for components subjected to mud and water.

Road Conditions & limitations of manufacturing techniques:

A new 2-wheeler was launched in the market by an Indian manufacturer in collaboration with a
foreign company. The design was very successful abroad, First lot of vehicles were imported from
overseas and things worked well for a short duration. Soon manufacturing was started in Indian
plant and then there were many warranty claims for chassis. It was because of poor manufacturing,
low quality material and also bad condition of roads in India. It was decided to redesign the chassis
as per Indian manufacturing standards, material and rough road conditions. CAE group carried out
FEA (Static, Dynamic and Fatigue) to improve life as per the new input data and modifications proved
successful.

Design Abuse: Design abuse means component is designed for some specific application / function
but its been used by the customer in a different or creative way. Clients in particular in Asian countries
are so clever that they can use the product in a manner which the design engineer might not have
thought about in his most creative and imaginative dreams!
27
Past, Present & Future of FEM

a. Tractor Fender (mudguard): Mudguard (fender) of tractor is designed for guarding the driver
against mud, but if you visit a village 3 to 4 people sitting on the mudguard is a quite common scène.
Tractor is not only used for agricultural application but also for transporting people from one place
to other. In developed countries the manufacture is bound to replace the failed component free of
cost only if its been used as per given instruction and not otherwise. But in Asia, could you imagine
a company saying village people, please do not use our tractor mudguards for sitting purpose or
instructions like one person per fender please!

b. Lassi Making (Washing) Machine: In Punjab people are fond of Lassie (specially prepared sweet
butter milk). One day in a hotel, group of 25 students arrived and ordered lassi. Hotels in Punjab
usually have a electircal special lassi making machine. But on that particular day hotel’s lassi machine
was out of order. Hotel owner, the street smart Sardar, took stock of the situation & thought manual
preparation would take lot of time & these young students will become impatieint & angry, but at the
same time it was not wise to loose such a big order. He saw the washing machine in the basement
& thought oh man! Washing machine also work similar to lassi machine then why not to prepare the
lassie using washing m/c it self. Sardarji ordered his staff to pour all the basic in gradient in appropriate
proportion in the washing machine. He started the machine and collected lassi from outlet meant
for soap water. Lassie was very very bubbly & testy. To his surprise the group of students immediately
ordered for another round & started visiting the hotel frequently. In short time the hotel became
so famous for lassi that people started visiting it from even other corner of the city. Neighboring
hoteliers got jealous and got the secrete by bribing the staff. Later every hotel in the region order the
same brand of washing machine & started using it as lassie making machine .

Washing machine is not designed and made for lassie preparation and imagine it might have got out
of order after some time (say within warranty period). The hotelier would call service engineer and
abused him in special words language! and demand for immediate replacement of failed machine,
without telling him the real reason for failure.

Now During the monthly warranty meetings all the departments will point their figure toward
design or R&D department, without actually thinking about the basic / root cause of failure. And the
responsibility of the analysis will be transferred to a poor CAE engineer like you and me, supposed to
solve the problem in a very short time. He will start from the basic drawing, generation of CAD model,
meshing & analysis without thinking about and taking in to account the actual manner in which the
machine has been used, and prepare and submit nice colorful report within the deadline ….

Failure analysis always consumes lot of time. Its responsibility of CAE engineer & CAE manager to
convince the R&D /Design chief that failure analysis requires more time & instead of starting the FEA
just for the sack of report submission, priority should be given to understand the problem first and
finding the root cause.

But at the same time there is no need to fear failure analysis problems. It’s the most challenging
type of job for any CAE engineer. The process of failure analysis and problem fixing teaches lot of
things that could not be learned by just doing routine meshing, BC’s and result type of activity using
software .

28
Practical Finite Element Analysis

Rather than increments, promotions or monetary benefits what gives most satisfaction to any CAE
engineer is to see components analyzed / designed by him getting fitted on thousands of vehicles and
doing well in the field.

2.6 Future of FEM

Research related to solution techniques & post processing tools is more or less saturated and in future
focus will be mainly towards reducing time, improving quality & flow pattern of mesh generated by
auto meshing. Also one can expect very powerful & user friendly commands for brick meshing (&
hopefully auto brick meshing too!).

At the moment Indian companies are competent with structural analysis. In near future demand will
be for high end analyses like NVH, CFD, Fatigue & Crash analysis.

29

Potrebbero piacerti anche