Developed Mathematical Model for Indeterminate Elements with Variable Inertia and Curved Elements with Constant Cross-Section

Issues such as analysis of indeterminate structural elements that have variable inertia as well as a curved shape still have no closed form solution and are considered one of the major problems faced by design engineers. One method to cope with these issues is by using suitable the ﬁnite element (FE) software for analyzing these types of elements. Although it saves time, utilization of FE programs still needs professional users and not all engineers are familiar with it. This paper has two main objectives; ﬁrst, to develop simple mathematical models for analyzing indeterminate structural elements with variable inertia and that have a curved shape with constant cross section, this model is much easier to be used by engineers compared to the FE model. For simplicity and saving time, a MATLAB program is developed based on investigated mathematical models. The force method combined with numerical integration technique is used to develop these models. The developed mathematical models are veriﬁed using the suitable FE software; good agreement was observed between the mathematical and the FE model. The second objective is to introduce a mathematical formula to determine the accurate number of divisions that would be used in the mathematical models. The study proves that the accuracy of analysis depends on the number of divisions used in the numerical integration. The optimum number of divisions is obtained by comparing the output results for both FE and developed mathematical models. The developed mathematical models show a good agreement with FE results with faster processing time and easier usage


Introduction
Beams with variable sections have a lot of practical applications in concrete and in steel structures.Tapered beams made of welded plate are an example of variable sections beams that are used in steel structures; these types of beams are used mainly in rafter and columns of frames or as continuous beams, and the main advantage of these type of section is they are economical where the beams have variable heights along their length so that the beam capacity equals the acting moment [1,2].In concrete structures, beams with variable sections are used in many structural applications and in many precast concrete buildings [3].
Designing and analysis of variable section beams are considered one of the problems that engineers face.Due to lack of knowledge, using the mean value of stiffness or ignoring variation of stiffness along element length are used in the design process of such elements, which is not accurate and can lead to progressive collapse of the structure.In some special structures that have elements with variable stiffness, it is essential to find an accurate analysis method for these elements.One of the main issues in designing a beam with variable inertia is that the section which is subjected to maximum stress need not necessarily be at the midspan of the beam due the variation in beam depth, which leads to changing the inertia along the beam length [4].
Many researchers were motivated to study the behaviour of beams with variable sections under different loading scenarios.Boiangiu et al. [5]; Ece et al. [6] and Felsoufi and Azrar [7] studied the dynamic behaviour and the free body vibration of beams with variable cross section.Yang et al. [8] and Dumitrache [9] studied the shear stress distribution and analysis for variable sections beam.Yang et al. [8] found that design codes do not accurately calculate the stress distribution in both the cracked and elastic stage using the effective shear force method.
Koo Lee et al. [10], Rojas and Ramirez [11], and Sapountzakis and Tsiatas [12] investigated the elastic linear behaviour of beams with variable sections using the classical beam theory, and others studied the nonlinear behaviour of such beams [13,14].However, most of the proposed design equations are still complicated to be studied by the engineers.
Another structural element that has a beautiful and attractive architecture view is arches.Arches were previously used by the Persian, Egyptian, Babylonian, Greek, and Islamic civilizations for different types of structures, but the ancient Romans were the first to use them as bridges.Arches have been used in some bridges in China and have been used widely in Islamic mosques, castles, and palaces.
Nowadays, arches are used in different structural applications as railway and roadway bridges, pedestrian bridges, or as part of a building, hangers with a large open space without intermediate columns.
e basic structural theory for arch is that it converts all stress into compressive and eliminates tensile stresses.is is useful especially for materials which carry less tensile stress.
Similar to beams with variable sections, engineers face difficulties in analyzing arches and determining the critical cross section and also calculating straining action along the arch length.Due to the importance of arches in our structures, many researches study the behaviour of arches under different loading conditions with different shape profiles.Sonavane [15] applied basic principles of the flexibility method in the analysis and design of symmetrical circular arches under different loading conditions.e results obtained from the analysis were compared with the finite element (FE) model results.MATLAB program was used to determine the internal forces in the circular arch design.
Ghannam and Najm [16] investigated the buckling behaviour of Islamic arches under different loading conditions.King and Brown [17] conducted the manufacturing and design of curved steel members using the British code (BS 5950-1 : 2000).King and Brown [17] included worked examples to clarify the design of arched steel used in different applications.
Designers tend to use FE software as one of the most effective methods that can be used in the analysis of arches or straight beams with variable cross sections.However, there is still no exact mathematical model for solving such problems.
Based on the above literature, it can be concluded that the main problem that faces engineers in designing structural elements that have variable inertia as well as that have a curved shape is the complication of the design methods that are available for these types of structures.Besides, not all engineers are familiar with using FE programs.erefore, the main aim of this paper is to cover that gap in this area and to provide simple mathematical analytical models that would simplify the analysis and design of structural elements that have variable inertia and a curved shape.In this paper, the flexibility and numerical integration methods are used to develop simplified mathematical analytical models.ese investigated models are used to evaluate the straining actions for straight beams with variable cross section and curved beams that have constant moment of inertia.To ease and save time of the solution process, simple MATLAB programs are established based on the investigated mathematical models.e output results from these models are verified using the suitable finite element software [18].It was found that there is a good agreement between the developed program and the result of the FE program.
By comparing the output results from both the FE software and developed mathematical models, it is observed that the number of divisions used in the numerical integration of the mathematical models play a main role to obtain accurate results.
erefore, the second objective of this paper is to present mathematical formulas for curved and straight beams with variable inertia to determine the optimum number of divisions used in the analytical model of each.A numerical analysis of the output results is applied to investigate these formulas.By adjusting the number of divisions used in the developed models, the output results show a good agreement with FE results.

Flexibility Method (Force Method)
e flexibility or force method was originally developed by James Clerk Maxwell in 1874. is method is very common and a powerful method for the analysis of statically indeterminate structures without limitations as in other structural analysis methods.
e method is based on transforming the structure into a statically determinate system and calculating the value of redundant forces required to restore the geometric boundary conditions of the original structure.e force method (also called the flexibility method or method of consistent deformation) is used to calculate reactions and internal forces in statically indeterminate structures.e method can be used for either plane or space structures subjected to loads or forced deformations.It is well known that the analysis of beams with variable stiffness along length in a closed form solution cannot be obtained.erefore, it is important to modify the existing method (flexibility method) for elements with constant stiffness to be suitable for beams with variable stiffness.Additionally, the exact analysis method for the curved beams with constant section is studied in this paper to fill the gap in the literature.
e flexibility and the stiffness for an element at a certain point and direction are reverse.e inverse of stiffness is flexibility.e relationship between flexibility and stiffness can be expressed as follows: where r ij is the stiffness matrix and δ ij is the flexibility matrix of an element.Flexibility matrix for an element is generally determined by the following equation: 2 Advances in Civil Engineering (2) Equation ( 2) can be simplified according to the type of structure.For beams, equation ( 2) can be simplified into the following equation, where the bending moment M x is the governing staining action for the beam cross section: where (M ix , M jx ) is the corresponding bending moment diagrams for force at ends (i, j), respectively, (E) is the modulus of elasticity, and (I x ) is the cross section moment of inertia.e solution of the numerical integration can be obtained using trapezoidal rules presented in the following equation: e equations of the force method can be written in the matrix form as follows: where f i and f j are the corresponding member force at the ends i and j, respectively, θ i and θ j are the member rotation at the ends i and j, respectively, and δ ij is the deflection in the direction of the i redundant due to a unit load in the direction of the j redundant.

Analytical Model for Beams with Variable Inertia.
is section describes the proposed analytical model through an application in which the internal moments at supports are determined for the beam due to clockwise unit rotation separately at both ends of the beam as shown in Figure 1.
Given θ i � 1.0 (rotation at each end of the beam separately), where h is the depth of the beam at the midspan, h x is the depth of the beam at distance x from the left side of the beam, and b is the beam width.
e equations of the force method: By transformation from x to ξ, δ ij equations take the form presented in the following equation: For the numerical integration of function Z i , assume the interval ξ is divided into n parts.e values of the function Z i are given in Table 1 using n � 8 parts.
e values of the integrals can then be calculated using equation (4), and the solution of the integrals is given below: Advances in Civil Engineering us, the force method equation in numbers and their solutions: e stiffness matrix r ij represents the stiffness matrix for a rod or beam with variable cross-section stiffness EI x .Figure 2 shows the bending moment diagrams corresponding to rod ends unit rotation.As shown in this figure, the values of bending moments differ significantly compared to the traditional method for constant stiffness.

Finite Element Model and Verification for the Analytical Model.
is section presents a case study which demonstrates the usage of the proposed model to solve a statically indeterminate beam with variable cross section.
e beam has a total length of 4 m and fixed end moments at both ends; the cross section at the first end is 0.25 × 0.5 m rectangular cross section and 0.25 × 0.3 m at the other end of the beam; and the modulus of elasticity was assumed to be 2.1E + 10 Pa.
e model was first simulated numerically using the finite element (FE) general software ABAQUS [18].e beam was simulated using solid element C3D8R, which is an 8-node linear brick with 3 translational degrees of freedom at each node [18].e model meshing is shown in Figure 3.
e meshing was used with maximum aspect ratio of 2; boundary condition at the beams end was applied to two reference points at beams ends.Each reference point is coupled with the beams end surface.
e loads applied were a rotation of 1 rad at each end.
e rotation was separately applied for each end in two different steps.e results and the staining actions produced from the rotation at each end is separated from the result of rotation at the other end (they are not accumulated or added together).
e results of the ABAQUS model is indicated in Table 2.It should be mentioned that the larger section is referred as End 1, the smaller section is referred as End 2, and the member end moment calculated by the FE model is referred as M FE .
A sensitive analysis has been done in order to investigate the effect of the number of division on the accuracy of the proposed model for different beam height ratio at both its ends.Figure 4 shows an example for the sensitive analysis in the case of height ratio at different ends being 0.6.In Figure 4, the vertical axis shows the ratio between the end moment obtained by the FE model (M FE ) and the end moment obtained by the proposed model (M model ).e horizontal axis shows the number of division used in the analysis, and the number of division used in this sensitive analysis is 2, 4, 6, 8, 1, and 12.It should be mentioned that the proposed model was programmed using the MATLAB program [19], which makes it easier to use the proposed model for different numbers of divisions.It can be observed the use of number of nine divisions gives a good result that is in reasonable agreement with the FE model results.
Table 2 shows a comparison between the result obtained by the FE model and that obtained by the proposed model.In Table 1, M FE refers to the end moment obtained by the FE model at both Ends 1 and 2 due to rotations � 1 rad at both ends separately, where M model refers to the end moment obtained by the proposed model using 8 divisions at Ends 1  Advances in Civil Engineering and 2. As can be seen from the table, there is a very good agreement between the FE and the proposed model, which shows that the proposed model can accurately predict the end moment of indeterminate beams with variable cross section.
Regression analysis was used to propose a simple equation to predict the required number of integral division (n) based on the ratio between beam depths at both ends (h 2 /h 1 ).e obtained number of divisions will give the best result that is in good agreement with the FE model; the proposed equation is shown in the following equation.
e R-squared value (R 2 ) for this equation is 0.99, which shows the accuracy of this model to predict the best number of integral divisions.
is is illustrated in Figure 5.
where n is the required number of integral divisions, h 1 is the height of the beam at the first end, and h 2 is the height of the bema at the second end where h 1 ≥ h 2 .Advances in Civil Engineering 5

Analytical Model for Curved Element with Constant
Inertia. is section describes the proposed analytical model through an application in which the internal moments at supports are determined for the circular arch due to clockwise unit rotation at both ends of the arch separately and a horizontal displacement at end (B) as shown in Figure 6.
Illustrative diagram for the solution using the proposed model is presented in Figure 7.
Step-by-step solution using the proposed model is indicated in equations ( 13)- (18).
Given z i � 1 (Z 1 and Z 2 are the rotation at both ends, respectively, and Z 3 is the horizontal displacement at End B), For the numerical integration of functions, the interval is divided into 8 pieces.Function values f i are given in Table 3.
e numerical integration was solved using the method of averages: e values of the integrals:  6 Advances in Civil Engineering e equations of the force method in the numbers: eir solution provides To construct a single diagram M zj , it is necessary to fix the unit diagrams M xj and put them together: Diagrams M zj are shown in Figure 8. Ordinates of the individual diagrams M xi , corrected diagrams M xi • X zj , and individual diagrams M zj are shown in Table 4.

Finite Element Model and Verification for the Circular Arch Analytical Model.
is section shows a case study, which verifies the accuracy of the proposed model to solve a statically indeterminate curved beam (arch) with constant cross section.e arch has a radius of 6 m and a height of 4 m with fixed end moments at both ends, the cross section of the arch was 0.25 × 0.4 m, and the modulus of elasticity was assumed to be 2.1E + 10 Pa.An outline of the arch is shown in Figure 9.
Similar to the previous case, the model is first numerically simulated using the finite element (FE) general software ABAQUS [18].e arch is simulated using solid element C3D8R.
A sensitive analysis has been done in order to investigate the effect of the number of divisions on the accuracy of the proposed model.e result of the sensitive analysis is presented in Figure 10.In Figure 10, the vertical axis shows the ratio between the end moment obtained by the FE model (M FE ) and the end moment obtained by the proposed model (M model ).e horizontal axis shows the number of divisions used in the analysis; the number of divisions used in this sensitive analysis is 4 to 16.Similar to the first case, the proposed model was programmed using the MATLAB program [19].It can be observed the use of number of divisions equal to twelve gives result that is in reasonable agreement with the FE model results.At the number of 12 divisions in Figure 10, the ratio of M FE /M model is equal to 0.96, 0.98, and 1.14 at End 1, middle, and End 2 of the arch, respectively, due to rotation at End 1. Due to horizontal displacement at End 2, the ratio of M FE /M model is equal to 1 and 0.95 at End 1 and   Advances in Civil Engineering middle of the arch, respectively.It should be noted that the solution is done faster using the developed program compared to the FE model.For instance, for structural element with variable section, the time needed to build and execute the FE model is 10 times that required to input parameters and get solution from the developed numerical model.Regression analysis was used to propose a simple equation to predict the required number of integral division (n) based on the ratio between arch height and its radius (f/r).e obtained number of division has good agreement with the FE model, and the proposed equation is shown in equation ( 16).
e R-squared value (R 2 ) for this equation is 0.98, which shows the good accuracy of this model to predict the best number of integral divisions.is is illustrated in Figure 11.(24)

Conclusions
From the litterateur review, it was concluded that the complication of the available design methods is considered the main problem that engineers face in designing structural elements having variable inertia as well as a curved shape; besides, not all engineers are familiar with using FE programs.e objective of this paper is to cover the gap in this area and provide simple mathematical analytical models that would simplify the designing and analysis of structural elements that have variable inertia and a curved shape.e developed program was obtained using the MATLAB software, and the output results from this program were verified using the FE software [18].e following conclusions can be drawn from this paper: (1) Significant proposed analytical model based on the flexibility method was developed for straight beams with variable inertia and curved beams (circular arch) with constant inertia.(2) Good agreement has been found between the proposed model and the numerical model which gives more confidence in using the proposed model in structural analysis for beams with variable cross sections and arch with constant cross section.Advances in Civil Engineering (5) New mathematical formulas were investigated for straight beams with variable inertia and curved beams with constant inertia to obtain the optimum number of divisions used in the proposed model.(6) e developed mathematical models for analyzing straight elements with variable inertia and a curved shape with constant inertia could be a helpful method used in teaching structural analysis of such elements for undergraduate students.(7) In the future research, the proposed model will be modified to be used with arched beam with variable cross sections.

Figure 1 :
Figure 1: e proposed model for solving beam with variable section.

Figure 2 :
Figure 2: Bending moment diagrams due to unit rotation at ends.

Figure 3 :
Figure 3: FE model developed by ABAQUS program.

Figure 4 :
Figure 4: Sensitivity analysis for number of division effect on the result of the end moments.End moment at (a) End 1 due to rotation at End 1 and (b) End 2 due to rotation at End 2.

Figure 5 :
Figure 5: Proposed equation for prediction the required number of integral divisions.

Figure 6 :
Figure 6: e statical system of the circular arch and the proposed main system.

Figure 7 :
Figure 7: Bending moment diagram due to redundant.

Figure 11 :
Figure 11: Proposed equation for predicting the required number of integral divisions.

Figure 9 :
Figure 9: FE model developed by ABAQUS program for circular arch.

Figure 10 :
Figure 10: Sensitive analysis for number of division effect on internal forces of a circular arch.(a) Internal forces due to rotation at End 1 and (b) internal forces due to horizontal displacement at End 2.

Table 1 :
Values of the function Z i using n � 8 parts.

Table 2 :
Comparison between the FE model results and the proposed model results.

Table 3 :
Function values f i at different sections on the arch.
3) A new program was developed using the MATLAB software based on the proposed model.e new program can analyze beams with variable sections and arches with constant section consuming less time compared to the ABAQUS program, which makes the proposed model more reliable compared to the other method of analysis.For structural element with variable sections, the time needed to build

Table 4 :
Function values M xi at different sections on the arch.
Advances in Civil Engineeringand execute the FE model is 10 times that required to input parameters and get solution from the developed numerical model.
(4) e number of divisions used in the investigated analytical models play a main role to get best agreement with FE solution.