Paper ID #14685

Using Finite Element Methods to Calculate the Deflection of an Orifice Plate

Subject to Uniform Pressure Distribution
Dr. Aneet Dharmavaram Narendranath, Michigan Technological University
Dr.Aneet Dharmavaram Narendranath is currently a Lecturer at Michigan Technological University (Michi-
gan Tech). He received a PhD in Mechanical Engineering-Engineering mechanics in 2013. Subsequently,
he worked as a visiting assistant professor at Michigan Tech from 2013-2014 and then as an Engineer
at the French Nuclear Commission (CEA) in France until 2015. His research interests are mathematical
modeling of fluid physics. His pedagogical interests are development of mathematically oriented coures
in mechanical engineering.
Mr. Prathamesh Prashant Deshpande, Michigan Technological University

Prathamesh Deshpande is pursuing his Doctor of Philosophy in Mechanical Engineering from Michi-
gan Technological University, Houghton, MI, USA. He received his undergraduate degree in Mechanical
Engineering in 2013 from Savitribai Phule Pune University (previously University of Pune), Pune, In-
dia. During the academic year 2014-2015, Prathamesh was granted his Master of Science degree from
Michigan Technological University with a Report option. His report was on the computational study of
a non-cylindrical, non-comformable CNG tank mounting on a pickup truck frame using finite element
methods. He is currently serving as a Graduate Teaching Assistant under Dr. Aneet Narendranath for a
senior level introductory finite element method course. The major duties as an assistant involve guiding
senior students in gaining computational knowledge of the applications of the finite element methods in
solving simple engineering problems.

Mr. Madhu Kolati, Michigan Technological University

Mr. Madhu Kolati is currently pursuing his doctoral studies in the field of acoustic meta-materials at
Michigan Technological University. His academic and research interests are in the areas of Solid mechan-
ics, Finite Element Methods, Elastodynamics and Phononic Crystals. Prior to joining Michigan Tech, he
worked as a Design Engineer at Caterpillar Inc.

Mr. Datta Sandesh Manjunath, Michigan Technological University

Mr. Datta Sandesh Manjunath, is currently pursuing his Masters in Mechanical Engineering at Michigan
Technological University. He has graduated from Amrita Vishwa Vidyapeetham, India with a B.Tech
degree in Mechanical Engineering. He is currently doing his report, on Impact analysis of a pick up truck
having a non-cylindrical, non-conformable CNG Tank using Finite Element Modelling. His academic and
research interests are in the areas of Solid Mechanics, Composite Materials and Finite Element Methods.
He also works as a Student Coach in the Engineering Learning Center. After graduation he seeks to work
as a Product Development Engineer in a reputed firm.

American
c Society for Engineering Education, 2016
 Using Finite Element Methods to Calculate the deflection of an
orifice plate subject to uniform pressure distribution

Abstract

As part of an elective course in Finite Element Methods (FEM) for senior level and graduate
students in mechanical engineering, an ASME standard for flow measurement devices is used to
design an orifice plate. Students are given a certain set of flow condition and equipment
constraints that they must adhere to. As part of the design process, they are required to evaluate
their orifice plate for strength via finite element methods and determine if the plate’s transverse
deflections due to uniformly distributed pressure are within set limits.
To design the orifice plate, a symbolic solver (Wolfram Mathematica) is used to solve the
governing fourth order differential equation of this problem (plate equation in polar coordinates),
with appropriate boundary conditions. Results from the symbolic solver are juxtaposed with
results from a GUI/Menu driven FEM package (Altair Hyperworks suite). Both the symbolic and
menu driven solutions are compared with each other and with published relationships.
Governing equations for bending of plates, in polar coordinates (for the orifice plate) have the
need to resolve mathematical singularities for “1/r”, for “r=0” type terms. This when reconciled
using symbolic solvers allows a better grasp of the esoteric inter-relationships between various
terms in the governing equations, which are akin to design variables. This allows students to use
this esoteric knowledge to better apply GUI/menu driven solvers for engineering design.
The primary pedagogical goal of this work allows the exertion of importance of governing
equation based modeling to improve a “behind the scenes” understanding of GUI/menu driven
FEM efforts. Students are made aware of the use of engineering standards and validation of
numerical solutions based on numerical accuracy and convergence of solution through
comparison with analytical data.
Introduction and Philosophy

Modern day FEM is closed attached to the advent of mathematical and matrix algebra methods in
the design of aeronautical structures 1,2,3,4 . Primarily, FEM is a method of approximations to solve
field problems. The power of FEM packages is realized when the fundamental field problems
governing the engineering design are “encompassed” in irregular shapes. In this paper, regular
shapes are: square/rectangular geometrics, circular cross sections.
In the combined undergraduate and graduate level mechanical engineering course on
“Introduction to Finite Element Methods”, the aim of this ongoing work is to temper the use of
commercial FEM packages with a sound understanding of fundamental engineering physics and
differential equations (classical analysis). This allows for an emphasis on the strengths and
limitations of the software package and the analytical solutions and methods to make good FEM
pre-processing decisions. An exposure to analytical methods also allows students to design
experiments/technology and to analyse and interpret results and data obtained effectively. To do
this, a project is introduced in designing an orifice plate (standard flow measuring device) through
the use of a commercial FEM package (the Hyperworks suite) with result validation obtained
from analytical solutions from the Theory of Elasticity (the Biharmonic equation is used).
Our university is an ABET accredited university. The exercise described in this paper is in
accordance with ABET’s 2015-2016 criteria for accrediting engineering programs. ABET’s
criteria states “The curriculum must require students to apply principles of engineering, basic
science, and mathematics (including multivariate calculus and differential equations); to model,
analyze, design, and realize physical systems, components or processes; and prepare students to
work professionally in either thermal or mechanical systems while requiring topics in each area.”
The following student outcomes are outlined by ABET in their criteria for accrediting engineering
programs and they will be referred to as ABET (a)-(k) as is customarily the case:
(a) An ability to apply knowledge of mathematics, science, and engineering
(b) An ability to design and conduct experiments, as well as to analyze and interpret data
(c) An ability to design a system, component, or process to meet desired needs within realistic
constraints such as economic, environmental, social, political, ethical, health and safety,
manufacturability, and sustainability
(d) an ability to function on multidisciplinary teams
(e) An ability to identify, formulate, and solve engineering problems
(f) An understanding of professional and ethical responsibility
(g) An ability to communicate effectively
(h) The broad education necessary to understand the impact of engineering solutions in a
global, economic, environmental, and societal context
(i) A recognition of the need for, and an ability to engage in life-long learning
(j) A knowledge of contemporary issues
(k) An ability to use the techniques, skills, and modern engineering tools necessary for
engineering practice.
The objectives (primary and supplemental), successes and failures of this study are described
briefly in the project assessment table 1.
 Objective Success Reason for Failure Comments
Use of analytical equations and Yes n/a n/a
symbolic solvers to solve the en-
gineering problem

Validation of results obtained Yes n/a n/a

from FEM package with analyt-
ical solution

Validation of results obtained No Limited experimental Students need to start acquir-

from FEM package with exper- data found by stu- ing data at earlier stage.
imental results dents.

Supplementary goal of commu- Yes n/a ASME style extended ab-

nication of scientific results stracts were authored by stu-
dents

Supplementary goal of identify- Yes n/a During this project, students

ing other engineering problems uncovered many applications
with same fundamental core (machine components) of the
annular plate and some pur-
sued these alternate design
problems.

Table 1: Project Assessment Table. Since this paper describes a work in progress, some supple-
mentary goals were added.

Organization of Paper

In the succeeding sections, we discuss the use and relevance of symbolic solvers and how it
allows for development of new skills by students. The evolution of the design project that
students are to complete and the design project itself are described. Sample results of FEM
package (Hyperworks Suite) driven analysis and symbolic solver (Wolfram Mathematica) are
described and compared. This discussion of results also includes a brief description of analytical
modeling that is an input to the symbolic solver. Finally, the pedagogic results with respect to
engineering simulation are discussed along with a brief history of the fundamental differential
equation (Biharmonic equation) at the core of the technology being designed in this.

Use of Symbolic Solvers

Symbolic solvers (also known as “Computer Algebra Systems”) automate the solution of systems
of linear equations through the interpretation of symbols (coefficients, independent variables etc.).
The application of symbolic solvers has had a dramatic, positive impact on science and
technology with various accounts of its effectiveness and application available 5,6,7,8 and its usage
in education as a primary or complementary portion to enhance student understanding of
concepts 9,10,11,12 . These solvers lend themselves exceptionally well to classical analysis of
engineering problems. In our FEM course, Wolfram Mathematica is chosen for classical analysis
because of its powerful differential equation solver, NDSolve. The author has tested higher order,
non-linear, stiff spatiotemporal differential equations in engineering analysis through the use
Mathematica’s NDSolve function and results have been highly favorable. The author has
focussed NDSolve on numerical or analytical solutions of non-linear partial differential equations
as part of his research experience.
Since symbolic solvers allow for the interpretation of both numbers and symbols, their versatility
is greater than traditional procedural programming techniques such as FORTRAN, C/C++. They
do possess some drawbacks in the learning curve associated with them and that they traditionally
use JAVA based interpreters unlike the compiler driven FORTRAN, C/C++. However, for the
solution of fundamentally important differential equations in elasticity, fluid mechanics, heat
transfer (among many fields), symbolic solvers allow for a greater ease in obtaining numerical or
analytical solutions than the more time-consuming procedural programming methods. It is this
niche that is explored by mechanical engineering students in this FEM course, through the
obtention of closed form analytical solutions using symbolic solvers.
Closed form analytical solutions allow for an intimate understanding of the relationship between
various terms in a differential equation and their weights and how these terms relate to the physics
being captured. The weights of differential equations are generally design (structural or fluid)
parameters or properties. Some classical examples are provided in equation 1 and equation
2.  2
∂ u ∂ 2u ∂ 2u

∂u q̇
=α 2
+ 2+ 2 + (1)
∂t ∂x ∂y ∂z ρcp

∂ 2u ∂ 2u
 
=c (2)
∂t2 ∂x2

The heat equation 1 results from an application of the conservation of energy and Fourier’s law of
conduction. The parameters α, ρ, cp are thermal properties of the material through which heat
propagation is being studied. These are essentially weights for the second order heat diffusion
term or the volumetric heat generation term. The magnitudes of these terms can change the nature
of the physics being studied by changing the weight of the second order-in-space effect vs the first
order-in-time effect.
The wave equation 2 is a result of Newton’s second law. The parameter c is the velocity of
propagation of a wave. In case this equation is applied to an axial prismatic bar element, then c is
a measure of a stress pulse that travels through the bar element. Again the value of this parameter
has an effect on the physics being studied.
Parametric studies on the heat equation, wave equation or other such equations that are closely
connected to mechanical engineering design may be easily performed in symbolic solvers.
Learning new skills

The symbolic solver, Wolfram Mathematica allows students to learn a new skill that they were not
exposed to previously in our mechanical engineering curriculum: functional programming. The
programming skills of engineering students in general curricula is that of “procedural
programming” such as that gained in MATLAB . R Functional programming for the symbolic
solution of differential equations is a highly efficient paradigm.
Functional programming through Mathematica allows for such differential equations to be solved
in fewer lines of code and in much quicker fashion than through procedural programming. One of
the main drawbacks of procedural programming as compared to functional programming is that
neophytes in programming have an over-reliance on iterative control statements such as “for” or
“while” loops or “if” statements which are inefficient due to their need for changing states of
variables and mutability of data. In contrast to procedural programming, functional programming
allows for easier application of functions than the use of iterative control statements for the
solution of differential equations.

Evolution of the idea behind the design project

During the FEM course, students are exposed to the Laplace equation (steady state form of
equation 1) and the Biharmonic equation (the square of the Laplace equation). The Laplace
equation and the Biharmonic equations are particularly relevant to mechanical engineering
students because of their applicability in “potential theory”. Potential theory is equally applicable
to fluid dynamics and heat transport and solid mechanics. Some prominent examples of potential
theory are “inviscid flow over an airfoil” (important problem in fluid dynamics) and the “stresses
in an axially loaded plate with a hole” (important problem in solid mechanics) to name a few.
This juxtaposition allows students to recognize the universality of partial differential equations to
areas of engineering.
In tandem, students study the application of matrix structural analysis for solid mechanics and
flow systems, simple cases of weighted residual methods for solid mechanics and fluid mechanics
systems and the use of the Hyperworks Suite. Students are exposed to Mathematica as as
symbolic solver to obtain analytical solutions to reduced governing equations. Once the students
have acquired adequate experience in both analytical methods and the FEM package, they are
required to complete a term project. The processing steps towards obtaining solutions to
engineering problems (field problems governed by partial differential equations) via the
Hyperworks suite and that through Mathematica are described in figure 1 and figure 2.
To perform a finite element analysis for an engineering problems, first time students of this course
are faced with the Hyperworks students’ manual. It is a veritable source of information. However,
the Hyperworks students’ manual describes pre-processing, solution and analysis
(post-processing) based strongly on “rules-of-thumb”. This can lead to confusion for students
who are not informed on the mathematical or continuum mechanics background that led to these
“rules-of-thumb”. Recognition of classical or fundamental equations that led to these
“rules-of-thumb” and their analytical solutions help students make pre-processing decisions when
they attempt to design engineering problems. The analytical solutions to classical problems can
be obtained by the solution of their respective partial differential equations in Mathematica with
significant ease.
The idea of this design project is to allow students to make explicit connections between
governing equations for classical problems to make informed decision on FEM steps when using
packages such as the Hyperworks suite.

Figure 1: Standard FEM Process as adapted from the Hyperworks suite students’ manual.

Figure 2: Sequence of steps used by Wolfram Mathematica’s NDSolve solver to solve differential
equations. As a solution, NDSolve returns an interpolating function over the domain of a differ-
ential equation when specified with boundary conditions. The required order of the interpolating
function may be specified as an argument to NDSolve. This interpolating function may be plotted
or manipulated to create interactive solutions as a means to post-process results

Description of Design Project

The term project is that of finding the maximum deflection (or radial deflection profile) of an
annular disk (a bidirectional orifice plate) that is fixed on its outer circumference and free on its
Figure 3: The original problem set-up: Annular disk with inner radius b and outer radius a sub-
jected to a uniform pressure force p0 . Its outer circumference is fixed while the inner circumference
is free to deflect.

inner circumference under the effect of a uniformly distributed traction force (pressure) on its
surface (figure 3). This problem is solved with the Hyperworks Suite (Hypermesh and MSC
Nastran). The maximum deflection (or radial deflection profile) obtained in the Hyperworks suite
needs to be validated by comparing with analytical or experimental results.
For validation of their FEM results, students need to solve the Biharmonic equation (the square of
the Laplace equation) to study the deflection of a thin plate. Students are made aware of the
singularity at the center for a disk without a hole and how they may overcome it through a
modification of boundary conditions or through the L’Hôpital’s rule. An annular disk, however,
does not suffer from this singularity and students are required to come to this conclusion
independently. Also of significant consequence is the application of boundary conditions for the
Biharmonic equation applied to the annular disk. Students need to apply “fixed” boundary
condition on the outer circumference and “free” boundary conditions (no shear force or bending
moment) at the inner circumference. Proper application of boundary conditions allows for a
spherically symmetric closed form solution to be obtained. Students are encouraged to arrive at
the boundary conditions through hand calculations and are recommended Mathematica to solve
the eventual Biharmonic equation. Further validation may be obtained by comparing the closed
form solution with solutions widely available in the literature.
An ASME Standard 13 that describes various types of orifice plate geometries, material
considerations, their utility, related measurement uncertainties and corresponding equations is
provided. The ASME standard’s utility is to allow students to create a part model with pressure
taps and other accessories that may be manufactured. Manufacturing an orifice plate and its
paraphernalia is not part of this course in FEM.
They are required to report their results as an “extended abstract” as per ASME conference
guidelines. Most students have not experience writing scientific articles at this juncture in their
career and the organized nature and conciseness of the “ASME extend abstract” format is a good
foil for them to practice their scientific communication skills. To recount, the multi-fold nature of
the project is as follows:
 • Use of Hypermesh to preprocess/set-up the problem, its boundary conditions, loading
conditions, solution and post-processing of results.
• Solution (with the use of Mathematica) to the analytical equation that describes this
problem (Biharmonic equation in polar coordinates) with appropriate boundary conditions.
Empirical (experimental) or analytical relations need also be sought to provide validation of
results. A multitude of analytical equations outnumber empirical equations for the
deflection of an annular plate subject to fixed boundary conditions at its outer
circumference and with a uniform pressure load. This abundance of analytical equations
persuaded students to use them. These are available in design data handbooks, textbooks
and classical journal publications.
• Communication of scientific results in a concise fashion in the format of an ASME
extended abstract.

Results

In this section, the following methods and results are discussed:

• The deflection of an orifice plate obtained by analytically solving the governing equation
(Biharmonic equation in polar coordinates). The general sequence of steps is shown in
figure 2.
• The deflection of the orifice plate as obtained in Hyperworks suite. The general sequence of
steps is shown in figure 1.
• The deflection of the orifice plate

Governing Equations (Biharmonic equation in polar coordinates)

The governing equation for the deflection of an annular plate (orifice plate) subject to uniform
pressure (traction force) on one of its faces is the two-dimensional Biharmonic equation in polar
coordinate (equation 3). This is a standard equation in plate theory that results when a force
balance is applied by considering bending moments and shear forces. The bending moments and
shear forces can be represented through the deflection w(r).

w0 (r) w00 (r) 0000 2w000 (r) p0

− + w (r) + = − (3)
r3 r2 r De

The radial Biharmonic equation is commonly denoted as shown in equation 4. In our work, we
have unravelled it into the form suggested by equation 3 for ease of input into Mathematica’s
NDSolve function and a means of term-by-term manipulation through the introduction of weights
if required.    
1 d d 1 d dw p0
r r =− (4)
r dr dr r dr dr De
In equation 3, the azimuthal component is neglected since the deflection is radial only. The
pressure is p0 while De is the flexural rigidity given by equation 5 wherein, the Young’s modulus
is Ey , h is the thickness of the plate and ν is the Poisson’s ratio of the material of the plate.
Standard assumptions of small strain and isotropic material are applied.

Ey h3
De = (5)
12 (1 − ν 2 )
The boundary conditions for the problem are so as to model the inner radius (r = b) of the
annular plate as being “free” while the outer radius (r = a) as being fixed. In terms of bending
moments and shear forces, the four boundary conditions applied to the Biharmonic equation,
equation 3 are:
w(a) = 0 (6)
w0 (a) = 0 (7)
w0 (b) w00 (b)
− 2 + w000 (b) + =0 (8)
b b
w0 (b)
+ w00 (b) = 0 (9)
b
The problem set-up in Mathematica’s NDSolve function is shown in figure 4.

Exact Solution of Governing Equation

There exists an exact solution of the Biharmonic equation subject to these boundary conditions as
given in equation 10

w(r) = c1 ln(r) + c2 r2 ln(r) + c3 r2 + c4 + w1 (10)

4
Where w1 = p0 r /(64De ). The coefficients c1 through c4 are also available but given their length
(multi-term numerator and denominator), they have been omitted. Obtaining the coefficients
requires the evaluation of the four boundary conditions as simultaneous equation. This is a
bookkeeping exercise which is not of the essence for the focus of this work. A detailed
methodology for similar situations of circular plates is described in literature 14,15,16,17 . It is
interesting to note that analytical accounts and approximate solutions show preponderance over
experimental data. This was realized independently by the students too and reflects in their choice
of mode of validation. Students overwhelmingly used published analytical equations to validate
their own analytical/Mathematica work and results obtained from the Hyperworks Suite.

Analytical equations

For loaded circular plates with a central hole, for a variety of boundary conditions, analytical
equations for the maximum deflection, ymax , of the plate are described by equation 11. These are
commonly tabulated 15,17 .
 In[10]:= a = 10 * 10 ^ - 3; (*Outer radius*)
b = 6 * 10 ^ - 3; (*Inner radius*)
ν = 1  3; (*Poisson's ratio*)
p0 = 0.1 * 10 ^ 6; (*Pressure force*)
Ey = 200 * 10 ^ 9; (*Youngs Modulus*)
h = 1 * 10 ^ - 3;(*Thickness of the annular disk*)
Ey h3
De = ; (*Flexural rigidity of disk*)
12 1 - ν2 

sol = NDSolve

(*Differential equation*)
w ''''[r] + 2  r w '''[r] - 1  r ^ 2 w ''[r] + 1  r ^ 3 w '[r] ⩵ - p0  De,
(*Fixed boundary condition at outer radius*)
w[a] ⩵ 0,
w '[a] ⩵ 0,
(*Free boundary conditions at inner radius*)
- Derivative[1][w][b]  b ^ 2 +
Derivative[2][w][b]  b + Derivative[3][w][b] ⩵ 0,
ν Derivative[1][w][b]  b + Derivative[2][w][b] ⩵ 0
,
w, {r, b, a}
(*Deflection plot vs radius*)
wplot = Plot[Evaluate[w[r]
Figure 4: NDSolve set-up of Annular /. sol],
disk {r, b,
subject to a},
uniform pressure with fixed outer circumfer-
ence and freePlotRange → {{0, a}, Automatic},
inner circumference BaseStyle
is solved using equation 4.
→ {FontWeight → "Bold", FontSize → 18},
AxesLabel → {"Radial location [m]", "Transverse deflection [m]"},
PlotStyle → {Black, Thick}, ImageSize → 800]
(*Export["/home/dnaneet/MEEM2911__MEP2/mma/orifice_plate_deflection.pdf",
4
p0 a
wplot,ImageResolution→600]*)ymax =k (11)
Evaluate[w[b] /. sol];
Ey h3
The coefficient k (stress at
(*Deflection inner radius*)
concentration factor) depends on the outer to inner radius ratio (a/b ratio)
and is tabulated in Timoshenko et al. 15 . In future sections, the phrase “tabulated equations” refers
Out[16]= w → InterpolatingFunction
Domain: {{0.006, 0.01}} 
to equation 11. Output: scalar

Comparison of solutions

Table 2 shows a comparison of maximum deflection of the orifice plate, at its free circumference,
as obtained by Mathematica’s NDSolve (mma), MSC Nastran (msc) and tabulated equations
(tab). These simulations were run by the instructor and the GTAs (graduate teaching assistants) as
a preliminary study. The outer radius is a and the inner radius is b. In all cases, the thickness of
the plate is 1 mm. The stress concentration factor k in the equation for deflection, ymax,emp was
obtained by linear interpolation for those a/b ratios that were not explicitly tabulated. The
following properties are used: Ey =200 GPa, p0 =0.1 MPa, ν=0.33. For these cases solved using
Printed by Wolfram Mathematica Student Edition
the Hyperworks Suite, the “Optistruct” solver is used with CQUAD4 (shell type quadrilateral
element with 4 nodes as per the Hyperworks Students’ manual) and CHEXA (Solid element of
Hexahedral type with 20 nodes as per the Hyperworks Students’ manual) for 2-D and 3-D
geometries respectively. Both 2-D and 3-D treatment revealed no difference in deflection values
obtained, as should be the case for this purely radial problem. The “Optistruct” solver is a
“Structural Analysis Solver (linear and non-linear); market-leading solution for structural design
and optimization” as per Hyperworks’ parent company “Altair Engineering”. Little detail is
available about the inner workings of “Optistruct”.
Table 3 shows a selection of student results. As noted in the table, a variety of element types are
available within the Hyperworks Suite and after grid independence is achieved, the solution
(deflection of annular plate) shows some variation with the element type.

Outer and inner radii ymax,mma [m] ymax,msc [m] ymax,tab [m]
a=5.0 mm, b=2.0 mm 3.09 × 10−8 3.64 × 10−8 2.93 × 10−8
a=5.0 mm, b=2.5 mm 1.79 × 10−8 2.10 × 10−8 1.79 × 10−8
a=5.0 mm, b=3.0 mm 8.31 × 10−9 11.1 × 10−9 8.96 × 10−9
a=5.0 mm, b=4.0 mm 6.11 × 10−10 13.7 × 10−9 6.21 × 10−10
Table 2: Comparison of maximum deflection of the annular plate as obtained through different
methods of solution.

Discussion

For an elective Finite Element Method course, students were exposed to the use of a general
purpose symbolic solver to use for analytical solutions in conjunction with traditional FEM
software (Hyperworks suite) to design an orifice plate. Most students who performed this opted
for a bi-directional orifice plate per the ASME Standard for flow measurement devices. An orifice
plate is installed with its outer circumference fixed and its plane subject to hydrostatic pressure
(modelled as a traction force). Due to the traction force, the orifice plate deflects (however in
minuscule fashion) and this deflection may be captured by both specialized FEM software and
general purpose symbolic solvers. The symbolic solver is used to solve the governing equation
(Biharmonic equation in radial coordinates) with appropriate boundary conditions to obtain
deflection (and radial deflection profile) of the free inner circumference of the orifice plate. A
closed form solution is also available for this problem and may be obtained using the symbolic
solver again.

Major Learning Points

The underlying philosophy that students were exposed to and many appreciated is the use of
fundamental governing equations and symbolic solvers to make FEM parameter and design
decisions. The distinctive nature of Mathematica allow for the solution of differential equations
(partial or ordinary, linear or non-linear) whilst focusing on the equation itself and not on
algebraic manipulation. Students appreciate the reduction of tedious algebra using symbolic
manipulation software. With less time spent on algebra, more time is available for students to
 Plate Pa- ymax,mma [m] ymax,msc [m] ymax,emp [m] Element type
rameters
a=100 mm, 0.2701 × 10−3 0.2699 × 10−3 0.2711 × 10−3 First order tetrahedral and
b=25 mm, first order hexahdedral
p0 =3.5 MPa, with the “Optistruct”
Ey =210 GPa, solver
h=1 mm
a=100 mm, 0.2199 × 10−3 0.2220 × 10−3 0.2190 × 10−3 Mixed element type with
b=50 mm, “Optistruct” solver
p0 =1.0 MPa,
Ey =210 GPa,
h=5 mm
a=80 mm, 0.2592 × 10−3 0.2077 × 10−3 0.2596 × 10−3 first order Tetrahedral ele-
b=20 mm, ment, “Optistruct” solver
p0 =1.0 MPa,
Ey =200 GPa,
h=2 mm
a=70 mm, 0.1157 × 10−3 0.1206 × 10−3 0.1160 × 10−3 Mixed element type with
b=30 mm, the “Optistruct” solver
p0 =0.5 MPa,
Ey =210 GPa,
h=3.5 mm

Table 3: Sample student data with ν=0.3. The element type, order, size were decided through
convergence of Hyperworks solution to analytical solution. In the Hyperworks suite, first order
tetrahedral elements are known as “Linear Tetra”, first order hexahedral elements are “linear Hex”.
The mixed element type is used a shell type element that has both triangular and quadrilateral
elements. Both 2-D and 3-D element types were used by students.

engage is higher-level synthesis and understanding of engineering implications. A model that

depicts Hyperworks suite juxtaposed with analytical solutions obtained from Mathematica is
shown in figure 5.
1. Solution accuracy and convergence: One of the important criteria that students based
their FEM solution accuracy was on the comparison of deflections with that obtained from
an analytical solution as obtained from Mathematica. This comparison allowed students to
make three critical decisions: (a) Accuracy of results obtained from FEM software (b)
choice of element type based on convergence of FEM solution to analytical solution
obtained from Mathematica and (c) concurrent grid dependency. These three factors
implicitly validate results obtained by the FEM software. The use of Mathematica to solve
the Biharmonic equation allows for an understanding on how changing certain parameters
(element type, number) can have an effect on simulation results and that analytical solutions
may be used to make these FEM decisions which directly impact engineering design.
2. Choice of element type: In the Hyperworks Suite, there exists a decision making paradigm
with little preamble on how to choose element types. A variation in choice of element type
is demonstrated by students as there are different element types that result in similar
 convergence. Both two-dimensional or three-dimensional element types are feasible for the
problem at hand, with differing degrees of convergence with the analytical solution. Having
an analytical solution obtained through Mathematica helps students arrive at this choice of
element type and near equality of most element types experimented with for this problem.
3. Convergence criteria: The Hyperworks student manual provides a simple example with a
cantilever with a point end load and compares the displacement as obtained from the
computational solution with an analytical solution from basic mechanics of materials. A
mesh dependence study is performed and it is observed that the deflection of the cantilever
from both methods is fantastically close (error<1%). However, this is not the case with
stresses with the error being nearly 5%. This difference is assigned to different
extrapolation methods for “in-element-stress” calculation. A number of students shows lack
of awareness of the choice of convergence criteria: should one use displacement or stress?
It would seem that displacement, which is the field variable being solved for by the
Biharmonic equation may be used as a convergence criterion with less complexity than
stress. However, students need to be made aware of this nuance. The utility of an analytical
equation and its solution helps with choice of convergence criteria in the Hyperworks suite
for this problem.
4. Circumventing of- or bolstering confidence on- “rule-of-thumb” modelling decisions:
Hyperworks Suite Students’ Manual provides useful “rules-of-thumb” in gauging the
convergence of results and accuracy of computational technique. However, no exposition is
provided on the fundamental equation being solved. These rules of thumb are particularly
useful if students were to understand an analytical approach. A good balance between
analytical solutions and computational solution techniques must be struck. The use of a
symbolic solver for a fundamental elasticity problem of gravity allows for this balance.
5. Ambiguity of application of boundary conditions: This item is also a response to the
preliminary review comments received on this paper about modelling of boundary
conditions. In Mathematica, free boundary conditions must be expressed explicitly through
equations for shear force and bending moment (equations 6-9). In case of the Hyperworks
suite, the equivalent treatment is the application of constraints to nodal degrees of freedom.
Fixed boundaries are applied with the 6 nodal degrees of freedom associated are set to zero.
The initial state is always a free boundary condition and this need not be explicitly applied.
In the Hyperworks suite, “free boundary conditions” are implicit unlike their application in
Mathematica.
Figure 6 shows the number of students who used and those who did not use analytical results as a
decision making tool for their analysi with the Hyperworks suite. Out of 42 students, 33 students
primarily used the analytical solution to the Biharmonic equation to make modelling decisions in
Hyperworks. In figure 7, of the 9 students who failed to recognize and follow this process, 8 used
the incorrect governing equations while only 1 did not have any governing equation.
Per Richard Felder 18 , “creating a course to achieve specified outcomes requires effort in three
domains: planning, instruction and assessement/evaluation”. It is apparent to the author that the
instructional process may need to include emphasis on time required to acquire publications. Had
students been warned sufficiently of lead times to acquire publications to validation of their
Figure 5: Juxtaposition of traditional FEM method as adapted from Hyperworks’ student manual
with analytical solutions from Mathematica (denoted by rounded rectangles with the double bor-
der). The analytical solutions provide critical information to the FEM package and allow for a
scientific approach to pre-processing and post-processing instead of a decision making paradigm
driven by “rules-of-thumb”.

results with experimental data, they may have had the opportunity to perform validation of their
numerical results not just with analytical and approximate solutions but also with experimental
results.
In the end-of-semester instructor evaluations, this method of juxtaposing Mathematica with the
Hyperworks suite received favorable comments. The author would like to include that this was a
first attempt at integrating the use of symbolic solvers and analytical solutions from theory of
elasticity to provide credence to package driven FEM simulation results. Previously, only
elementary mechanics of materials concepts were utilized in the validation of simple problems
such as those demonstrated by bending of beams and axial deflection of bars. In the future, more
emphasis will be given to differential equation/theory of elasticity based modelling.
The current course-level assessment tool of a project report in the form of an “ASME Extended
abstract” is planned on being retained. However, this project report did not go through a peer
review process and was critiqued traditionally by the instructor only. In future iterations of this
process, other modes such as mid-project oral presentation of results to peers will be
considered.

Brief Note on History of the Biharmonic Equation

The biharmonic equation provides solution to several physical problems concerning bending of
clamped thin elastic isotropic plates, equilibrium of an elastic body under conditions of plane
strain or plane stress, or creeping flow of a viscous incompressible fluid. The solution to the
 A(33)

B(9)

Figure 6: Breakdown of efficacy of governing equations and symbolic solution in making informed
modelling decisions. A: Students who relied on analytical solution to make modelling decision. B:
Students who did not make accurate modelling decision due to incorrect or no analytical solution
use.

B1(8)

B2(1)

Figure 7: Breakdown of efficacy of governing equations and symbolic solution in making informed
modelling decisions. B1: Subset of students from category B (figure 6) who used incorrect ana-
lytical equations. B2: Subset of students from category B who had no analytical equations in their
final report.

biharmonic equation is a rich field with luminaries such as Poincaré, Picard and Painlevé (French
mathematicians with pioneering contributions in both pure and applied mathematics) once servied
as “Judges” for the prix vaillant (literal translation: Prize for Courage) worth 4000 Francs in the
mid 1800s.
Over the last ∼150 years, a great number of mathematical methods have been developed to obtain
the solution of the biharmonic equation. These methods and solutions have been used to define
and design many facets of technological progress. For more information on the Biharmonic
equation’s rich history of the Biharmonic equation, the reader is directed to Meleshko 19 .
A few results from students

Some of the student results are shown in figure 8 to 10. There were a variety of combination of
parameters that students focused on depending on their individual interests and motivation. The
core of each individual student result was still the comparison of results obtained via FEM with
analytical results and the use of symbolic solvers to help make FEM decisions.

Figure 8: Assessment of effect of annular disk (as a washer) thickness with a = 1.5”, b = 0.78”,
h = 0.1” to 0.18”, E = 28 × 106 psi and ν = 0.3. Contour plots show deflection (left) and Von
Mises stress (right)
Figure 9: Assessment of effect of element type and size on deflection obtained via FEM and
comparison with analytical solution. The annular disk has a = 250 mm, b = 50 mm, h = 15 mm,
E = 210 GPa and ν = 0.3. Contour plots show deflection (left) and bending stress (right)
Figure 10: Assessment of effect of element size on deflection obtained via FEM and comparison
with analytical solution. The annular disk has a = 10 cm, b = 5 cm, h = 0.5 cm, E = 210 GPa
and ν = 0.3. Contour plot shows deflection.
Responses to reviewer comments

The authors thank the program/session chair and the reviewers for their prompt comments and
wish to reflect on the comments received.
Reviewer 1 mentions that the word “bookkeeping” is “one word and not two”. This has been
corrected by the authors. (S)he also mentions that “The first sentence in the third paragraph is
patently untrue”. This is obfuscating since it is not clear to the authors whether the reviewer is
referring to the first sentence of the third paragraph in the abstract or that in the section on
“Introduction and Philosophy”. Nevertheless, we examine both and make a change to the first
sentence of the third paragraph in the abstract:
• “Problems in polar coordinates are complex in comparison to those in rectilinear
coordinates. This is because of the requirement of symmetry boundary conditions or the
neglect of constants of integration to avoid singularity type of errors when solving polar
coordinate problems analytically.” has been changed to “Governing equations for bending
of plates, in polar coordinates (for the orifice plate) have the need to resolve mathematical
singularities for “1/r”, for “r=0” type terms.”
• Our university is indeed ABET accredited and this project does address most of the ABET
student learning goals.
The authors would like to point out that their initial statement of polar coordinates being difficult
as compared to cartesian coordinates is not untrue. On performing cursory searches for
publications on this matter, it is revealed that students and practicing engineers have less
practice 20,21 with polar coordinates. There is also evidence presented 22 that students gravitate
towards Cartesian coordinates rather than use polar coordinates, even for problems that would be
amenable to analysis via polar coordinates. Intrinsically, this shows the apprehension that
students have towards polar coordinates. There are numerous publications that may be found that
show the preference of choice of coordinate system and the perceived difficulties therein.
Errors in spelling, errors in grammar have been corrected to the best of the authors’ abilities.
Refinement of statements through a different choice of phrase also have been made. This log of
changes is described in table 4.
Change made Reason for change
“we discuss the use and relevance of symbolic solvers and how this allows for Grammar
development of new skills by students.” changed to “we discuss the use and
relevance of symbolic solvers and how it allows for development of new skills by
students.”

“intepretation” changed to “interpretation” Error in spelling

“intepreters” changed to “interpreters” Error in spelling

“obtention of analytical solutions” changed to “obtention of closed form analyti- Clarification to maintain consistence in next statement that
cal solutions” used “closed form”

“Closed form analytical solutions allow for an intimate understanding of the re- Provides clarification to why analytical solutions are useful
lationship of various terms in a differential equation and their weights” changed and important
to“Closed form analytical solutions allow for an intimate understanding of the re-
lationship between various terms in a differential equation and their weights how
these terms relate to the physics being captured”
 “The symbolic solver, Wolfram Mathematica allows students to learn a new skill Changed “:” to “viz.”
that they were not exposed to previously in our mechanical engineering curricu-
lum: functional programming” “changed to The symbolic solver, Wolfram Math-
ematica allows students to learn a new skill that they were not exposed to previ-
ously in our mechanical engineering curriculum, viz., functional programming”

“Once the student has acquired adequate experience” changed to “Once the stu- Grammar
dents have acquired adequate experience”

“go through the application of matrix structural analysis” changed to “study the Avoiding the use of “phrasal verbs”
application of matrix structural analysis”

“The result obtained for maximum deflection (or radial deflection profile) needs
to be validated and this is where symbolic solvers come into the picture.” changed
to “The maximum deflection (or radial deflection profile) obtained in the Hyper-
works suite needs to be validated by comparing with analytical or experimental
results”

“They are required to report their results as an “extended abstract”” changed to Replacement of pronoun “they” with the noun“students”
“Students are required to report their results as an “extended abstract”

“unraveled” changed to “unravelled” Spelling variant as per the Oxford English dictionary

“GTAs” changed to “GTAs (graduate teaching assistants)” Clarification of acronym

“per the Hyperworks Students’ manual” changed to “as per the Hyperworks Stu- “as per” is appropriate since it means “in accordance with”
dents’ manual” in the Oxford English dictionary. “per” is used to denote
“for each”.

Addition of text: “The “Optistruct” solver is a “Structural Analysis Solver (linear Definition for “Optistruct”
and non-linear); market-leading solution for structural design and optimization”
as per Hyperworks’ parent company “Altair Engineering”. Little detail is avail-
able about the inner workings of “Optistruct”.”

“Miniscule” changed to “Minuscule” Error in spelling

“Modeled” and “modeling” changed to “Modelled” and “modelling” Error in spelling

“convegence” changed to “convergence” Error in spelling

“Hyperwork’s” changed to “Hyperworks” The proper noun is “Hyperworks”

“with esoteric information with little preamble” changed to “with little preamble” Omission of the phrase “with esoteric information”

“A variation in choice of element type is demonstrated by students as there are Clarification provided
many options” changed to “A variation in choice of element type is demonstrated
by students as there are different element types that result in similar convergence”

“it is seen” changed to “it is observed” Refinement of statement

“A lot of students showed significant confusion on” changed to “A number of Refinement of statement
students shows lack of awareness of”

“complication and its provenance” changed to “nuance” Refinement of statement

Table 4: Log of changes.

Reviewer 2 makes the following objective recommendations:

• I am unfamiliar with the NDSolve component in Mathematica. Would the author consider
including a few screenshots of example problems, just to provide a flavor for how a simple
equation might be solved?
• The paper desperately needs a figure of the problem in question: the orifice plate. Included
 in the figure should be the loading and boundary conditions.
• A figure illustrating sample results from a student’s finite element analysis (e.g. a stress or
deflection plot) would also be helpful.
The authors realize that these recommendations by reviewer 2 are important and have done their
best to incorporate these through figures 3, 4, 8, 9, and 10.

Acknowledgement

This elective course in “finite element methods” was conducted at Michigan Technological
University (Michigan Tech). The first author thanks Ms. Akanksha Laxman Hande, a graduate
student at Michigan Tech, for her input. The first author also thanks the consistent and rigorous
involvement of the graduate teaching assistants for this course, Mr. Madhu Kolati and
Mr. Prathamesh Deshpande.

References

[1] Alexander Hrennikoff. Solution of problems of elasticity by the framework method. Journal of applied
mechanics, 8(4):169–175, 1941.
[2] Robert J Melosh. Basis for derivation of matrices for the direct stiffness method. AIAA Journal, 1(7):
1631–1637, 1963.
[3] James L Tocher and Billy J Hartz. Higher-order finite element for plane stress. Journal of the Engineering
Mechanics Division, 93(4):149–172, 1967.
[4] Irons Ergatoudis, BM Irons, and OC Zienkiewicz. Curved, isoparametric,“quadrilateral” elements for finite
element analysis. International Journal of Solids and Structures, 4(1):31–42, 1968.
[5] FC Mbakogu and MN Pavlovic. Some applications of computer algebra to classical problems in plate theory.
International Journal of Mechanical Engineering Education, 27(4):347–365, 1999.
[6] MN Pavlović. Symbolic computation in structural engineering. Computers & structures, 81(22):2121–2136,
2003.
[7] Mustapha Er-Riani and Olivier Sero-Guillaume. Shapes of liquid drops obtained using symbolic computation.
Journal of Symbolic Computation, 40(6):1340–1360, 2005.
[8] Robert M. Corless and David J. Jeffrey. Solution of a hydrodynamic lubrication problem with maple. Journal
of Symbolic Computation, 9(4):503–513, 1990.
[9] Radian Belu and Alexandru Belu. Teaching physics with computer algebra systems. In American Society for
Engineering Education. American Society for Engineering Education, 2009.
[10] Radian G Belu, Irina Nicoleta Ciobanescu Husanu, and Alexandru Catalin Belu. Teaching mechanics with
maple. In American Society for Engineering Education. American Society for Engineering Education, 2011.
[11] JJ Rencis and HT Grandin. Mechanics of materials: an introductory course with integration of theory, analysis,
verification and design. In 2005 American Society for Engineering Education Annual Conference & Exposition,
pages 12–15, 2005.
[12] RE Link and SM Miner. Complementary usage of mathematica and i-deas in mechanism design. age, 4:1.
[13] ASME. Asme-mfc-3m-2004: Measurement of fluid flow in pipes using orifice. 2004.
[14] JJ Cao and AJ Bell. General solutions for a circular flat plate with a central hole loaded by transverse force or
bending moment uniformly distributed along a circumference. International journal of pressure vessels and
piping, 51(2):155–173, 1992.
[15] Stephen Timoshenko, Sergius Woinowsky-Krieger, and S Woinowsky-Krieger. Theory of plates and shells,
volume 2. McGraw-hill New York, 1959.
[16] Rudolph Szilard. Theory and analysis of plates. 1974.
[17] AM Wahl and G Lobo. Stresses and deflections in flat circular plates with central holes. Trans. ASME, 52(1),
1930.
[18] Richard M Felder and Rebecca Brent. Designing and teaching courses to satisfy the abet engineering criteria.
JOURNAL OF ENGINEERING EDUCATION-WASHINGTON-, 92(1):7–26, 2003.
[19] VV Meleshko. Selected topics in the history of the two-dimensional biharmonic problem. Applied Mechanics
Reviews, 56(1):33–85, 2003.
[20] Mike Ellis, Brian Williams, Habib Sadid, Ken W Bosworth, and Larry Stout. Math usage by practicing
engineers: What does it mean to curriculum planners? age, 9:1, 2004.
[21] Laurens Bollen, Paul van Kampen, and Mieke De Cock. Students’ difficulties with vector calculus in
electrodynamics. Physical Review Special Topics-Physics Education Research, 11(2):020129, 2015.
[22] Eleanor C Sayre and Michael C Wittmann. Plasticity of intermediate mechanics students’ coordinate system
choice. Physical Review Special Topics-Physics Education Research, 4(2):020105, 2008.