Attribution Non-Commercial (BY-NC)

6 views

Attribution Non-Commercial (BY-NC)

- Crystal Cathedral
- Cathedral Architecture
- The Cathedral
- Aeronautical Syllabus-Shivaji University- R.H.B. Ramamurthy
- Static & Dynamic analysis of piping system
- ejemplo de reporte de flow simulation
- Syllabus for First Year Second Sem
- Fluid k2opt
- Chap 6
- EFFECT OF VARIABLE VISCOSITY AND THERMAL CONDUCTIVITY OF MICROPOLAR FLUID IN A POROUS CHANNEL IN PRESENCE OF MAGNETIC FIELD
- Fluid Mechanics Syllabus
- Civil Engg. Scheme & Syllabus
- Mixed EHL Lubrication
- Nonlinear Convection towards an Exponentially Shrinking Sheet with Magnetic Field
- Fluid Dynamics
- WIND TUNNEL DESIGNS, Ahmed.pdf
- 1.Flow of Fluids
- 13904_FULLTEXT
- PDF
- Reynolds Average Navier-Stokes Equation

You are on page 1of 29

Alexey I. Borovkov Igor B. Voinov Sergey A. Sidorov Computational Mechanics Laboratory, St.Petersburg State Polytechnical University, Russia

Abstract The paper presents the results of a finite element (FE) analysis of the stress-strain state (SSS) of the Peterand-Paul Cathedral spire metalware that is one of the basic architectural and historical dominants of St.Petersburg (Russia) (Figure 1). This research has been fulfilled in view of restoration works carried out to a 300-anniversary date of the St-Petersburg foundation.

Mathematical and 3D FE beam models of the spire containing all geometrical features of a design such as 34 tiers, 4 beams cross-section types varying with the altitude of the tiers have been developed. For verification of the developed models, the comparative analysis of the natural frequencies of the spire metalware was implemented. This analysis was carried out on the basis of FE simulation with the natural frequencies, received by the experiments earlier carried out, was implemented. The analysis of overflowing the spire cross-sections by a wind flow was carried out with the use of the ANSYS/FLOTRAN software package FE. For mathematical simulating the processes of air overflowing, a model of incompressible liquid flow based on the stationary 2D Navier-Stokes equations, Reynolds-averaged, was used. The airflow turbulence was taken into account by the use of a semi-empirical differential k- turbulence models. On the basis of the FE solutions of aerodynamics problems the distribution of the wind pressure along the cross section contour and a spire height was defined. The analysis of 3D SSS of the beam spire metalware under the action of the computed wind pressure was implemented. The stress concentration area and the most dangerous construction fragments were stated. For obtaining a more precise local stress distribution in these fragments a multilevel submodeling method was applied. The following developed models were used: 1. The initial model 3D beam macromodel, 2. The first-level submodel 3D FE model of the most dangerous fragment, containing its geometrical features: connecting bolts, contact interaction, 3. The second-level submodel meant for defining local stress fields in the investigated fragment in the case of one-bolt loss, studying the local stress redistribution and determining a safety factor. All researches of SSS of the Peter-and-Paul Cathedral spire have been carried out with the application of the FE simulation ANSYS/Mechanical software.

Introduction

The cathedral of Saint Apostles Peter and Paul located in the fortress named after them (Peter-and-Paul Fortress), which had been founded as far back as during the reign of Peter the First and not once rebuilt since that time, still remains the most remarkable and beautiful symbol of the city of Saint-Petersburg. The greatest architects, engineers and mechanicians Domenico Trezzinij (Trezini), F.B. Rastrelli, D.I. Jourawski (D.I. Zhuravsky), L. Euler and many others participated in its building and subsequent modernizations. The Peter-and-Paul Cathedral was laid after the foundation of the Peter-and-Paul Fortress on 30 May of the year 1703. Before long, the rebuilding of the Cathedral was contemplated, and in 1714 it was commissioned to Domenico Trezzinij. He had to replace a wooden construction of the Cathedral by a stone one. In 1722 a stone belfry of the Peter-and-Paul Cathedral was completed, and, according to the conception of Peter the First and architect Trezzinij, it became one of the main symbols of the new Russian Empire. In 1857, D.I. Jourawski, a talented military engineer, suggested that the wooden spire, which had required frequent repairs and renovations, should be replaced by a metallic framework, unprecedented in engineering practice till that time, based on the principle of light and durable bridge-girders. The same appearance of the Peter-and-Paul Cathedral remains today the architect Trezzinijs stone part of the Cathedral, the Djyakovs and Rastrellis stone belfry and the engineer Jourawskis metallic spire crowned with an apple and a weather-vane in the shape of angel implemented by the sculptor R. Zalemans design and rotating around the cross. The main goal of the research considered in the present paper was the development of a mathematical model of the metalware of the Peter-and-Paul Cathedral spire with high degree of adequacy to the unique historico-architectural structure, intended for the detailed investigation of the three-dimensional stressstrain state (3D SSS) of the spire metalware, exposed to the action of wind pressure. It should be mentioned particularly that the development of a mathematical model of the spires metalware was caused by a series of problems arisen during the process of the Peter-and-Paul Cathedral restoration, carried out to meet Saint-Petersburgs tercentenary. These specific problems are connected with a problem

of more precise definition of safety factors, which were chosen in designing and creating the Peter-andPaul Cathedral metal spire by D.I. Jourawski in the middle of nineteenth century, with the analysis of the influence of corrosion action lasting for almost 150 years, and with the detailed analysis of local weakenings (damages) of spire metalware components, i.e. losses of some bolts and rivets.

The Peter-and-Paul Cathedral spire consists of a metallic beam-like framework and a gilded copper sheath. The spires beam-like framework presents a 34-tier structure (Figure 2). Figure 3 demonstrates a typical appearance of a spire tier. The height of the beam-like framework taken from the spire basis first belt to the last, 35th tier is 40 m.

Figure 3. Typical appearance of a spire tier The spires shape presents a truncated octahedral pyramid with regular octagon in the horizontal cross section [6]. Figure 4 demonstrates one of 35 spire belts, a typical belt (collar) being of octagonal shape.

So, the Peter-and-Paul Cathedral spire consists of the following constructive components: a) Main rafter collar. It is composed of eight parts interjointed at butt-jointing ends with the use of cleats and bolts (Figure 4). The main collar is arranged on the stonework of the Cathedral belfry, at the 64 maltitude from the ground surface. b) Rafter ribs. Each of eight rafter ribs presents a pair of beams with angle-shaped cross section, united in such a way that their collective cross section acquires T-shaped appearance. c) Collars. Each collar (or belt) presents eight angle-shaped cross section beams connected with each other at their ends. The connection of these belt beams with almost vertical rafters is performed with the use of rivets, the upper head of which is of conical shape. Collar parts are strengthened by tension members made of sheet iron: around spire bottom with cross section 4 inch (10.16 cm 1.59 cm), then with cross section 3 inch (7.62 cm 0.32 cm) and, finally, with cross section 2 inch (5.08 cm 0.32 cm) [6]. It should be mentioned that these tension members are installed only up to the 27th spire belt. d) Diagonal struts. Diagonal struts were produced from sheet iron with dovetails at their ends, intended for fastening diagonal struts with ribs and collars [6]. Typical dimensions of all cross sections of beam-like components diminish with the increase of belt location altitude taken from the spire base main rafter collar. The height of each of 34 tiers also changes depending on tier location altitude over the spire base the higher the location of a tier, the less is its height. The most loaded areas in the whole spire metal construction are connections of neighboring belt rafters. In the most loaded lower part of the spire, a slant of a rafter with regard to the vertical spire axis is different at different tiers. Special components straps are used for their joining. The photograph of a typical connection joint of vertical rafters of neighboring tiers with belt beams, diagonal struts and straps is presented in Figure 5.

The second tier plays a special role in the spire metalware. Special articulated fastening joints (lugs with three-inch round pivot bolt) are attached to all rafters of this tier, which form the maximum angle with the vertical axis. Vertical tension bar performed in the shape of two iron sheets with rectangular cross section are fixed to articulated joints (Figure 6). These open bonds are sunk vertically downwards and fastened to massive metallic beams built in the stonework, lower than the belfry location, where the clock is installed. The tension of open bonds provides the steadiness of the spire, as a whole, against the overturning action of winds.

Figure 6. Tension bar The lower belt of the spire metalware is mounted on a massive stone base and rigidly fastened to it with the use of anchor bolts. The stone base presents a most upper part of the Peter-and-Paul Cathedral belfry. A spoke, that is a vertical round-cross-section beam whose diameter decreases with height, is attached to the upper part of the spire. The lower end of the spoke is rigidly built in the upper part of the spire and fixed by the use of iron wedges.

A copper gilded hollow ball (so-called apple) is attached to the upper part of the spoke. A revolvable device with the weather-vane Angel is installed in the central part of the spoke. A gilded copper cross is fastened to the upper part of the spoke. In contrast to the Angel, the apple and the cross are immovable parts of the construction. During the period of restoration work, it was established that all the components of the spire construction (rafters, belt beams, joints, diagonal struts as well as tension bar) were produced from the metal with the following properties: Youngs modulus 200 MPa, Poissons ratio 0.3, density 7800 kg/m3, yield strength 294 MPa.

Development of mathematical and finite element models for the spire metalware

The building of the three-dimensional model of the Peter-and-Paul Cathedral spire was carried out on the basis of drafts and drawings presented by the authorities of the State Museum of Saint-Petersburgs History, located inside the Peter-and-Paul Fortress. For detailed investigation of three-dimensional stress-strain state of the beam-like metal construction exposed to action of wind pressure, a modern efficient digital method were used the finite element method [10] and the ANSYS FE-software [12]. In order to build a mostly precise three-dimensional model of the spire metalware, there were taken into account all geometrical peculiarities of the beam-like framework, namely geometrical features of each belt and each tier of the construction. The three-dimensional beam-like model of the spire contains all principal beam components: rafter ribs of T-shaped cross section; collars consisting of belt beams of angleshaped cross-section; tension members; diagonal struts with dovetails at the ends, serving for fastening diagonal struts with ribs and collars, as well as open bonds tension bar. Simulation of tension bars was restricted to their part, which is located inside the spire metalware, from the basic rafter collar up to the third collar (belt). The three-dimensional beam model of the spire considers the lessening of typical dimensions of beam-like component cross sections depending on the rise of belt location altitude. For building a three-dimensional model, beam-like, three-node finite elements BEAM 183 [11] were used, each of which contained 18 degrees of freedom. The developed 3D FE model of the spire metalware (Figure 2) consists of 15043 finite elements (NE), 43828 nodes (NN) and contains 262968 degrees of freedom (NDF). Figure 7 presents a beam-like FE model of the first three tiers. This part of the three-dimensional model of the spire metalware is the most representative, inasmuch as it consists of the greatest number of various beam-like components: rafter ribs, collars, tension members, diagonal struts, open bonds drawing rods, as well as rigid, transverse H-shaped beams. The other peculiarity of this fragment of the spire metalware is the rafter slant alteration at the third belt level.

Figure 7. FE model of the first three tiers Figure 8 demonstrates photographs of the internal structure of the multiple-stage metalware of the Peterand-Paul Cathedral spire shot from the 21st belt altitude, from inside the spire and from below upwards; Figure 9 shows the corresponding fragment of the three-dimensional beam-like FE model of the spire metalware.

Figure 8. Photograph of the internal structure of the multiple-stage metalware of the Peter and Paul Cathedral spire

Figure 9. Fragment of the three-dimensional beam-like FE model of 25th and 26th tiers of the spire metalware

The photographs of the 25th and 26th tiers of the spire metalware are presented in Figure 10, and the corresponding fragment of the three-dimensional beam-like FE model is shown in Figure 11.

Figure 10. Photograph of the 25th and 26th tiers of the spire metalware

Figure 11. Fragment of the three-dimensional beam-like FE model of the 25th and 26th tiers of the spire metalware

Procedure

Verification of the developed mathematical and finite element models of the spire metalware

Aiming at the verification of the developed mathematical and finite element models, the analysis of natural frequencies and natural oscillation shapes applied to the Peter-and-Paul Cathedral spire was carried out. The values of natural frequencies obtained with the use of FE simulation were compared with those obtained experimentally, as a result of the research implemented in October, 1991, by scientists of the State Design Institute Lenproyektstalkonstruktsiya and described in the paper [26]. At the testing period, the investigated metalware did not contain the weather-vane Angel, which had been disassembled earlier and dispatched for restoration. All other constructive and architectural components of the spire metalware were presented in full volume. According to the results of the experiments performed, the natural frequency of spire oscillations occurred to be equal to f1exp = 0.917 Hz. The research of natural frequencies of the spire oscillations was carried out with account of the metalware sheath which, introducing additional distributed mass, affects natural frequencies. Mass introduced by the spire sheath is simulated by increasing material density of the beam-like construction. Thickness of gilded copper sheets of the sheath is about 3 mm, total mass of the sheath is equal to 8,455 kg, total side surface area being equal to Stotal = 378.5 m2, total volume being equal to Vtotal = 0.95 m3. Mass of the beam-like spire metalware comes to mspire = 47424 kg at the spire volume value Vspire = 6.08 m3. The density of copper is c = 8900 kg/m3. Hence, the total mass of the spire metalware together with the sheath is equal to M = mspite + msheath = 55800 kg. Thus, the equivalent material density exceeds density of copper by about 3.26% and is equal to eq = 9190 kg/m3. Besides, the three-dimensional beam-like model of the spire metalware was complemented by the models of the ball (Apple) and the spoke, which pending rigidity is significantly less than that of the spire itself. The cross located at the upper part of the spoke was simulated by concentrated mass on the spoke end. The lower spire belt as well as the tension bar are assumed to be built-in. For more accurate correspondence of the three-dimensional FE model of the spire metalware with the real construction experimentally investigated, mass of four persons, present inside the ball while performing experiments, was also taken into account. The analysis of natural frequencies was carried out not taking the Angel into consideration, because it had been absent in the metalware during the period of experiments. The Peter-and-Paul Cathedral spire was simulated by 3D 20-node finite elements SOLID 95, each of which contained 60 degrees of freedom [10, 11]. For the ball (apple) simulation, 8-node shell finite elements SHELL 93 were used, each of which containing 48 degrees of freedom [10, 11]. The following parameters were used for building three-dimensional FE model of the Peter-and-Paul Cathedral spire metalware; containing beam, shell and 3D finite elements: NE = 24,704 total number of FE; NDF = 358,800 total number of degrees of freedom. The three-dimensional FE model of the spire used for determination of natural frequencies is presented in Figure 12. Figure 13 demonstrates the fragment of FE model for the spire, the apple, and the spoke.

Figure 12. The three-dimensional FE model of the spire used for determination of natural frequencies

Figure 13. Fragment of FE model for the spire, the apple, and the spoke FE simulation with the use of the Lanczos method [11, 27] allowed determination of three lowest natural frequencies of the Peter-and-Paul Cathedral spire metalware: f1 = 0.92 Hz, f2 = 1.37 Hz, f3 = 3.89 Hz. According to the results obtained in [26], the first natural frequency defined experimentally is equal to f1exp = 0.917 Hz. Sufficiently good correspondence of computed (f1 = 0.92 Hz) with experimentally determined (f1exp = 0.917 Hz) values of first natural frequency gives convincing evidence that the developed mathematical and finite element models describe rigidity and inertia characteristics with the high degree of accuracy.

It should be noted that the main external factor having an influence upon the behavior of the spire metalware is wind pressure. Therefore, one of the main problems considered in this paper is the determination of this pressure. While defining a static constituent of wind pressure acting upon the investigated metalware, the normative force of the wind was taken as a main parameter of wind conditions of a region. Its value for a given geographical region is determined on the basis of a statistical analysis of climatological data for wind speed in this region. The A.I. Voeikov Chief Geophysical Observatory had developed a map of wind speeds of various provision rates, in which the territory of the former Soviet Union was divided in seven regions. (The map was named The regioning of the USSR territory according to wind pressure). It should be marked that speeds presented for each region correspond to their values at the 10 m-altitude in the condition of open, unprotected space, and conforms to two-minute averaging. Taking into consideration that Saint-Petersburg (Russia) is located in the second wind region, maximum normative wind pressure for which is qII = 0.30 kPa = 30.61 kg/m2, the value of normative wind pressure could be defined equal to uII = 22.18 m/sec (~80 km/h). It should be noted that the wind flow influence on the pyramidal spire construction leads to substantial nonuniformity of the distribution of external wind pressure along the spire belts contour. Therefore, from the research view, it is very interesting to implement calculation of exact distribution of external wind pressure on the octahedral side surface of the pyramidal spire, using the modern CFD-approach (Computational Fluid Dynamics, see, for instance, [18]). Such substantially nonuniform distribution of wind pressure will affect a breaking-away type of air current, turbulization of air torrent, initiation of negative pressure areas (the phenomena of inflow or rarefaction) and thereby fully conform to relationships of aerodynamics [19-22]. Hydrodynamics relations, based on the continuous medium model and obtained for incompressible liquids, occur to be correct for gases (in particular for air) as well. The measure of compressibility or alteration of moving medium density is Machs number M (E. Mach, 1838-1916) equal to the ratio of moving gas speed u to spreading speed of sound in it. It is useful to mention that at M < 1 the air flow is called subsonic, at M = 1 sonic, at 5 M > 1 supersonic, at M > 5 hypersonic. As a rule, air compressibility is taken into account if M > 0.3, corresponding to moving air speeds u > 100 m/sec. At Mach number M < 0.14, alteration of density and possible error in determination of air pressure by the formulae for incompressible liquids do not exceed 1%. Spreading speed of sound in atmospheric air is equal to 360 m/sec; hence, at air moving speed u < 50 m/sec (precisely this upper restriction of possible values of wind speed is typical for Saint-Petersburg) laws for movement of gas and liquid may be considered the same, with high degree of accuracy. As it is known, dimensionless Reynoldss number (criterion) is used in hydromechanics as a universal criterion allowing determination of liquid motion mode for various cases. Reynoldss number characterizes relationship between inertia forces and viscosity forces: Re = ud/v = ud/, allowing combined consideration of typical flow speed u (m/sec), e.g. u speed of undisturbed flow running on a body; typical linear dimension d (e.g. diameter of tube, length or transverse size of a overflowed body); kinematical coefficient of viscosity for liquids (m2/sec); liquid density (kg/m3); dynamic coefficient of viscosity for liquids (Pasec). Let us estimate Reynoldss number for the problem of overflowing the lower part of the spire by air current. Assuming spire diameter equal to d = 3.2 m, speed of running-on, air current equal to u = 20 m/sec and kinematical air viscosity at 15C equal to = 1510-6 m2/sec, we will get: Respire = 4.26106. Elevated constructions of cylindrical, conical or pyramidal shape (the Peter-and-Paul spire presented octahedral pyramid) are related to the class of poorly streamlined bodies.

When these bodies are overflowed by air current, due to friction, in air originates so-called speed boundary layer [21], flow speed in which quickly falls down to zero at body surface. The phenomenon of breaking-away of boundary layer is typical for badly streamlined bodies (e.g. cylinder, ball, parallelepiped etc). This phenomenon leads to emergence (in rear parts of these bodies) of dead zones (areas of slow flowing), in which distribution of pressure differs from that taking place in the case of overflowing by ideal (nonviscous) liquid. The phenomenon of breaking away of boundary layer also leads to formation of eddies and great loses of energy at rear parts of poorly streamlined bodies, that causes significant hydraulic resistance to overflowing such bodies. The point of boundary layers break away from body surface is generally characterized by the angular coordinate (00<<1800, =00 direction of running-on current). Aiming at determination of wind pressure on the side surface of the spire, a series of problems of overflowing various cross sections of the spire was solved (for metalware belts arranged at various altitudes and having different diameters). All computations were carried out with the use of the finite element analysis software ANSYS/FLOTRAN [12] intended for simulation and research of liquid and gas flow. As a series of model problems, preceding detailed investigation of overflowing the octahedral pyramid of the Peter-and-Paul Cathedral spire, there were solved the problems of overflowing an endless round cylinder for various Reynoldss numbers Re. For mathematical simulation of the processes of overflowing by air current, a model of incompressible liquid was used, on the basis of stationary two-dimensional Navier-Stokes equations averaged by Reynolds. Turbulent viscosity was determined with the use of semiempirical differential turbulence model k- model [31]. Notice that within the Reynoldss number range 103105 the currents break-away was laminar-like, although the track was turbulent. The break-away point of laminar current was disposed at = 80 87. At the values of Reynoldss number Re > 105, the flow in boundary layer became turbulent, and the breakaway point migrated downwards, along the current, to the rear region ( = 110) at the same time, the track region significantly shortened in comparison with the case of the laminar current break-away. In the case of overflowing the cylinder with diameter equal to d = 3.2 m by air current with u = 19 m/sec, Reynoldss number takes the value: Respire = 4106. Using this value, it is possible to estimate the thickness of boundary layer = d / (Respire) = 1.7 mm. For obtaining correct distribution of pressure p, that is a function only of longitudinal coordinate x in the plane boundary layer, and speed u(y) distribution with acceptable accuracy, meshes containing 7 finite elements in the boundary layer were used. Typical distribution of pressure over cylinder surface (air current is running on from the right to the left) is presented in Figure 14 Similar wind pressure distributions obtained on the basis of experimental researches are presented in literature sufficiently comprehensively (see, for instance, [13, 16]). As it is in the region of break-away laminar current, static pressure in the turbulent current region is almost invariable. Figure 15 demonstrates typical distributions of speeds u(y) in 5 points of the cylinder surface (air current is running from the right to the left). One may see a point on the cylinder surface, where

u y

y =0

= 0. This is the

point of the breaking-away of the currents boundary layer. It should also be mentioned that behind the break-away point in Figure 15 may be seen a discontinuity point at some distance from the wall, the condition

2u = 0 takes place speed function has the discontinuity point and, as a sequence, the current y 2

Figure 14. Typical distribution of pressure over cylinder surface (air current is running from right to left)

Figure 15. Typical distributions of speeds u(y) in 5 points of the cylinder surface (air current is running from right to left) Figure 16 presents a typical FE model of air space around the spire a barrier with octahedral cross section. The corresponding fragment of FE model is shown in Figure 17 Lines of current overflowing the spire (a barrier with octahedral cross section) are depicted in Figure 18; behind the spire, a reciprocal-

circulating zone is seen. Figure 19 presents a drawing of pressure distribution upon one half of the spire belt surface. It contains distinctly exposed edge effects, being a typical diagram of wind pressure distribution, e.g., on a gabble roof of a building. Just as in the case of a cylinder, static pressure on the spire surfaces in the turbulent current zone is almost invariable.

Figure 19. Drawing of pressure distribution upon one half of the spire belt surface Owing to the FE solutions of the air overflowing of different spire parts with different cross sections located at different altitudes, pressure distributions on different spire belts were obtained. Fundamentally nonuniform wind pressure on the spires side surface was thereby established and determined in full volume.

Fastening conditions

For correct statement of the general problem, the problem of connecting the basic spire rafter collar with the belfry stonework is of extreme importance. In the D.I. Jourawskis report, it is marked that rafters are fastened to the stonework with eight iron bonds placed inside the stonework and eight open bands (tension bar) fastened to massive iron beams built in the wall, below the space for the clock. Bonds arranged inside the stonework present a beam about 6.03 cm in diameter, the lower end of which, passing through a granite stone, terminates with a thread and a nut intended for tightening the bonds. Open or external bonds present two iron bands 8.89 cm 2.54 cm in size, ending at the upper part with lugs, that combined with a threeinch round pivot bolt, fasten them to two sheets surrounding ribs and fixed on the ribs with 10 inch (25.4 cm) bolts. [6] Considering the foregoing detailed description of the method of fastening the spire metalware to the stonework, one may assume than the basic collar of the spire rafters is absolutely rigidly fastened to the Peter-and-Paul Cathedral stone belfry.

Finite element research of three-dimensional stress-strain state of the spire metalware under action of wind pressure

In the analysis of three-dimensional stress-strain state of the spire metalware, we shall consider only the beam-like metal framework and neglect accounting the spire copper sheath, assuming that its influence on the rigidity of the whole construction is unessential. We also shall not take into account geometrical features of the spoke, the ball (Apple), the weather-vane Angel and the cross. Inasmuch as the simulation of the ball, the weather-vane Angel and the cross was not carried out in the stress-strain computations of the spire, the loads acting on the ball, the weather-vane and the cross were applied as equivalent load to the upper belt of the spire. Designed pressure acting on the angel and the cross was defined in accordance with the Engineering Norms established in RF and was taken p = 1.88 kPa. The windward area of the Angel is equal to Sangel = 5.1 m2, the windward area of the cross is equal to Scross 1.5 m2. Consequently, the designed wind action, expressed in equivalent loads, is: Pangel 9.59 kN for the Angel and Pcross 2.82 kN for the cross. Considering that the spire base (basic rafter collar) is rigidly fastened with the massive stonework of the belfry with the use of anchor bolts deepened into the wall, the spire base was assumed to be built-in; tension bars were also supposed to be built-in at the level of the lower belt of the first tier. Apart from the wind pressure, all the components of the spire are under the action of constant load own weight of the construction, that was also taken into account in stress-strain computations. Finite element research of the three-dimensional stress-strain state of the spire metalware was carried out with the use of ANSYS-software. For solution of large sparse systems of finite element equations (NDF = 358 800), both direct methods (e.g., frontal method [28]) and highly productive iterative methods of conjugated gradients were used with the preliminary improvement of conditionality [29, 30]. Let us carry out the analysis of the stress-strain state of the beam-like metalware of the Peter-and-Paul Cathedral spire exposed to a action of distributed forces on its surface, computed with account of overflowing and break-away current. With this aim in view, let us estimate the normal stress intensity field or field of equivalent stresses by Von Mises: i = ( 3 2 ss), where s tensor of 2nd rank, called also stress deviator, ss double compression of stress deviators for each of construction components. Figure 20 demonstrates the stress (i) intensity field for spire metalware, while Figure 21 presents stress (i) intensity field for four lower tiers of the spire.

Figure 21. Stress (i) intensity field for four lower tiers of the spire From the results obtained, one may conclude the following: the most stressed areas of the investigated construction are zones of fastening beams of the third tier as well as rafters and diagonal struts of the second and third tiers. At the same time, maximum Von Mises equivalent stress in these zones does not exceed imax = 35 MPa. Normal stress (i) intensities for other components of the beam-like framework are significantly lower. Taking yield strength for a maximum permissible stress, we define safety factor, which is equal to k = 8.4. It is necessary to notice that finite element computations of the three-dimensional stress-strain state of the beam-like spire metalware were implemented using solely beam finite elements. But it is principally impossible to describe correctly the local stress state of a connecting block of the spire metalware only using beam finite elements. Therefore, more accurate analysis of the stress state of the connecting block of metalware beams was carried out with the use of multilevel submodeling method.

Finite element research of the local three-dimensional stress-strain state of the spire metalware component with account of bolt connections, contact interaction and possible damages Multilevel submodeling method

For researches of the local three-dimensional stress-strain state of separate blocks of the spire beam-like construction, it is necessary to use the multilevel submodeling method based on the locality principle of the composite structure mechanics [32] which was developed in detail at the laboratory Computational Mechanics of S-Pb SPU and successfully tested in the process of solving various complex problems of structure mechanics. In the multilevel submodeling method, after performing a solution of a problem for a macromodel (at macrolevel i = 0), at each subsequent level i (i > 1) three-dimensional problems for the submodel of the i-level are subjected to solution. These problems may be problems of elasticity theory, elasto-plasticity theory, contact interaction mechanics, composite structure mechanics and, finally, threedimensional problems of mechanics of contact elasto-plastic interaction of composite structures depending on peculiarities of local stress state of that particular zone, where more precise finite element solution is desirable. Extracting submodel of i (i > 1)-level from submodel of (i-1)-level, it is necessary for conjugated surfaces of submodels S (i-1, i) to specify kinematics boundary conditions, which are the interpolated values of displacements obtained by solving the problem for submodel of (i-1)-level. Such specification of kinematics boundary conditions introduces some edge-effect-type disarrangement into local stress-strain state of submodel of i-level, but this disarrangement appears only in direct nearness to conjugation surfaces. If conjugation surfaces are located sufficiently far from concentration zone the interesting for us stress, and if the area where coincidence of the results obtained for submodel of i-level and submodel of (i-1)-level (U(i) U(i-1)) is definitely present, then, with the use of the multilevel submodeling method, may obtain correct results, even for problems with singular stresses (for example, problems with concentrated forces, mechanics problems for laminated composites with free edge effects, mechanics problems of interfacial cracks, mechanics problems of free edge delaminations) [33, 34].

For the most stressed block of the spire metalware, a three-dimensional model was built considering geometrical features of connection of beam-like components of the spire metalware (Figure 5). Submodel of the 1st level is intended for transition from beam model to three-dimensional model and more precise determination of local stress state. Figure 22 demonstrates the three-dimensional macromodel of the spire metalware and the submodel of the 1st level for the spire connecting block (this block is depicted in Fig. 5). The submodel of the 1st level is characterized by the following parameters: NES(1) = 32 769; NNS(1) = 97 144; NDFS(1) = 291 432. The photograph of the metalware fragment shot from inside the spire is presented in Figure 23, while Figure 24 demonstrates the three-dimensional model of the corresponding metalware fragment.

Figure 23. Photograph of the metalware fragment from inside the spire

For correct formulation of the kinematics boundary conditions, the three-dimensional submodel of the 1st level included beams (Figure 22), analogous to those used for building the macromodel. Boundary conditions obtained from the solution of the macroproblem were applied to beam butts. Kinematics boundary conditions (displacements and deflection angles) were transferred from beam finite elements to three-dimensional ones with the use of connection equations [35], linking the displacements and deflection angles at beam nodes adjacent to conjugation surfaces with the displacements of conjugation surfaces nodes (in this case, it was supposed that all nodes of conjugation surfaces belong to one and the same plane). Bolt straining was simulated by special finite elements; there were modeled bolts strained with 125 Nm-momentum. Computations were carried out with the use of the built up FE submodel of the 1st level. Figure 25 demonstrates the field of Von Mises equivalent stresses i. It is seen distinctly that computed equivalent stresses i do not exceed 60 MPa, while the yield strength value is: y = 294 MPa. Hence, safety factor of the considered spire metalware component comes to: kS(1) = 4.9.

Figure 25. Scheme of the sequential submodeling method. Forming submodel of the 2nd level

Multilevel submodeling

For more detailed analysis of the local stress-strain state of the spire metalware block, a three-dimensional FE submodel of the 2nd level (Figure 25) was built characterized by the following parameters: NES(2) = 19 068; NNS(2) = 54 077; NDFS(2) = 162 231. Using this submodel, the research of stress-strain state of the spire metalware block was carried out, contact interaction between adjacent components of the bolt connection being taken into account. 631 contact finite elements CONTACT 174 were used for simulating contact interaction. The necessity of the multilevel submodeling application was caused by extreme complexity of computations and, correspondingly, a long time required for their implementation. Thus, for example, the fulfillment of the research of multiple contact interaction between adjacent parts of the spire metalware block with the use of submodel of the 1st level would require vast computer resources and computation time (the required time for a Pentium 4-class computer, with 1 GHz-frequency and 1 Gboperating memory, would be about one week of uninterrupted work).

The results of FE computations are depicted in Figure 25 in the form of the field of Von Mises equivalent stresses obtained for the submodel of the 2nd level. It is remarkable that, just as in the case of the 1st level, normal stress (i) tensity does not exceed 60 MPa. Consequently, one may conclude that, in spite of redistribution of local stresses, safety factor of the considered spire metalware component remains unchanging kS(2) = 4.9.

It is of practical interest to know how the local stress-strain state of the spire metalware component would change in the situation when one fastening bolt is absent. With this goal in view, the computation of this situation, using the submodel of the 2nd level (with one bolt absent), was carried out. The results obtained are presented in Figure 26 in the form of the field of Von Mises equivalent stresses i, from where it is seen that Von Mises equivalent stresses do not exceed 62 MPa. This makes it possible to conclude that for the regular work of the given construction block it is sufficient to have two bolts in operation. Figure 27 demonstrates redistribution of intensity of local stresses i (Pa) in the neighboring bolts for the case of one bolt fell out of a typical connection. The result obtained show that if one bolt is lost (absent), the normal stress i intensity in the middle bolt increases from 31.3 to 34.4 MPa, i.e. by about 10%.

Figure 26. Redistribution of Von Mises equivalent stresses i in the case one-loss bolt

Figure 27. Redistribution of intensity of local stresses i (Pa) in the neighboring bolts for the case of one bolt fell out of a typical connection

As it was stated above, to provide regular work of the loaded spire metalware block, two operating bolts are enough. It is interesting to determine the least quantity of bolts necessary for the given block. For computations of bolts and rivet connections it is routine in engineering practice to fulfill analyses on shear and bearing strength. On the basis of assuming uniform distribution of tangential stresses for the case of shear strength and nonlinear pressure distribution law for the case of bearing strength, there are known formulae [38] for computation of necessary number of rivets, with the use of data on shear and bearing strength. By the proper averaging a the stress tensor components and using conventional structure mechanics formulae, it is possible to state that for the block under consideration, it is sufficient to insert in it only one rivet instead of three installed.

Conclusion

This paper demonstrates wide possibilities of using modern multipurpose computer technologies in solution of complex problems of mechanics. Solution of many problems is often impossible without using direct computer simulation. The accomplished work demonstrates the use of modern computer technologies for the analysis of the three-dimensional stress-strain state of the beam-like metal framework of the Peter-andPaul Cathedral spire exposed to the action of wind pressure. The problems solution was performed with the use of the ANSYS finite element software system. Significant attention was paid to the analysis of natural frequencies of the beam-like framework of the Peter-and-Paul Cathedral. The analysis clearly demonstrated the possibility to solve not only complicated static problems but to research dynamic features of constructions as well. Good coincidence of the computed value of first natural frequency and that obtained experimentally gives evidence of a high degree of adequacy of built-up mathematical and finite element models. It should particularly be marked the application of CFD-approach for determination of equivalent wind load acting on the metalware. Application of this approach allowed the obtaining of exact pressure distribution on the side surfaces of the pyramidal spire, taking account of all aerodynamic peculiarities: break-away currents, reciprocal-circulating motion, negative pressure areas etc. Analysis of the three-dimensional stress-strain state showed that the value of maximum stress, originated as a sequence of wind pressure, is several times less than the permissible stress level. A safety factor was defined, both on the basis of beam-like macromodel, which value comes to 8.4, and on the basis of

submodels of the first and second levels, the value of which in this case occurred to be equal to 4.9. The defined value of safety factor gives evidence of high strength and reliability of the construction. It should also be mentioned that the work demonstrates the multilevel submodeling method developed in detail in the laboratory Computational mechanics. This method allowed not only the studying in detail of real local stress distribution in the metalware connecting block, but also the analyzing of local stress redistribution in the case of the damage in the block caused by the loss of one of the bolts.

References

1. Borovkov A.I. Palmov V.A. Applied mechanics and reconstruction of the Peter and Paul cathedral broach, Part 1. Construction of the broach and Angel-weathercock. Trans. of SPbSPU. 2003. 1. 1947 p. (in Russian) Borovkov A.I., Gimmelman V.G., Palmov V.A., Voinov I.B., Mikhalenko A.E., Sidorov S.A., Applied mechanics and reconstruction of the Peter and Paul cathedral broach. Part 2. Finite element stress analysis of the metallic broach. Trans. of SPbSPU. 2003. 2. 7-27 p. (in Russian) Jourawski (Zhuravsky) D.I. Building art and practical mechanics. Description of works of building top part of the Peter-and-Paul Cathedral bell tower in St-Petersburg fortress and updating Cathedrals building that have been fulfilled in 1857 and 1858 years. Part I. Introduction. St.Petersburg 1858. (in Russian) Pilyavsky V.I. Peter and Paul fortress. Ser. Architecture of Leningrad. L.; M. (in Russian) Jourawski (Zhuravsky) D.I. Peter and Paul Cathedral spire. Communication Department Journal. S.-Petersburg 1860. (in Russian) Jourawski (Zhuravsky) D.I. Building art and practical mechanics. Description of works of building top part of the Peter-and-Paul Cathedral bell tower in St-Petersburg fortress and updating Cathedrals building that have been fulfilled in 1857 and 1858 years. Part II. Study of project for building top part of the bell tower. Part III. Determination of stresses of the iron rafters elements of the spire. St.-Petersburg 1858. (in Russian) Gorev V.V. Metal construction: In 3 vol. M.: Visshaya shkola, 20012002. Vol. 1.: Element construction. 2001; Vol. 2: Construction buildings. 2002; Vol. 3: Special constructions and buildings. 2002. (in Russian) Bernshtein S.A. Studies of structural mechanics history. .: Goststroyizdat, 1957. (in Russian) Timoshenko S.P. History of Strength of Materials. With a brief account of the history of theory of elasticity and theory of structures. USA Dover publications, Inc., New York. 1957.

2.

3.

4. 5. 6.

7.

8. 9.

10. Zienkiewicz O.C., Taylor R.L. The Finite Element Method. 5th edition. Vol. 1. The Basis. 2000. 689 P.; Vol. 2. Solid Mechanics. 2000. 459 p.; Vol. 3. Fluid Dynamics. 2000. 334 p. 11. Bathe K.-J. Finite Element Procedures in Engineering Analysis. Englewood Cliffs. Prentice-Hall. 1982. 12. ANSYS Theory Reference. ANSYS Inc., Canonsburg, PA. USA. Eleventh edition. ANSYS Release 5.6. November 1999. 1286 p. 13. SNIP. Building regulations. Loads and actions. Moscow., 2002. (in Russian) 14. Retter E.I. Wind Load on Construction. CRI industrial constructions, M.L.; 1936. (in Russian) 15. Barshtein M.F. Dynamic Analysis of the High Construction Exposed to Wind Action. Construction mechanic and design calculation. 1974. 6. (in Russian) 16. Korenev B.G, Rabinovich I.M. Dynamic Analysis of the Buildings and Constructions. Moscow.: Stroiizdat, 1984. (in Russian) 17. Petrov A.A. Design Calculation Exposed to Intensive Wind Actions. .: , 1998. (in Russian)

18. Anderson D.A., Tannehill J.C., Pletcher R.H. Computational Fluid Mechanics And Heat Transfer. Hemisphere Publishing Corporation, New York. 1984. 19. Prandtl L. Hydro- und Aeromechanic. Berlin, 1929. P. 175-208. 20. Loitsyansky, L.G., Fluid and Gas Mechanics. Moscow; Nauka Publ., (in Russian) 21. Chang P.K. Separation of Flow, Pergamon Press, 1966. 22. Van Dyke . An Album of Fluid Motion. Department of Mechanical Engineering. Stanford University. The Parabolic press. Stanford California 1982. 23. Kolmogorov. A.N. Local turbulence Structure in Incompressible Fluid for Very Large Reynolds Number. Report AS USSR. 1941. V. 30, 4. (in Russian) 24. Przemieniecki J.S. Theory of Matrix Structural Analysis. New-York: McGraw-Hill, 1968. 25. Yokoyama T. Vibrations of a Handing Timoshenko Beam Under Gravity. Journal of Sound and Vibrations. 1990. Vol. 141, 2, P. 245258. 26. Rasha I.K. Dynamic testing of Peter-and-Paul Cathedral Spire: State Design Institute Report Lenproectstalconstructsiya SPb., 1992. 31 p. (in Russian) 27. Grimes R.G., Lewis J.G., Simon H.D. A Shifted Block Lanczos Algorithm for Solving Sparse Symmetric Generalized Eigenproblems. SIAM Journal Matrix Analysis Applications. 1994. Vol. 15(1). P. 228272. 28. Irons B.M. A Frontal Solution Program for Finite Element Analysis. Int. Journal for Numerical Methods in Engineering. 1970. Vol. 2, 1. P. 523. 29. Incomplete Cholesky Conjugate Gradient (ICCG). Internally developed, unpublished work. ANSYS Inc. USA. 30. Preconditioned Conjugate Gradient (PCG) Solver. License from Computational Application and System Integration Inc. Illinois. USA. 31. Launder B.E., Spalding D.B. The Numerical Computation of Turbulent Flows. Computer Methods in Applied Mechanics and Engineering. 1974. Vol. 3. P. 269289. 32. Borovkov A.I., Palmov V.A. Locality principle in mechanics of composite structures. Preprints 3rd Int. Workshop Nondestructive Testing and Computer Simulations in Science and Engineering (NDTCS'99). St.Petersburg. Russia. 1999. H6-H7. 33. Borovkov A.I., Palmov V.A., Banichuk N.V., Stein E., et al. Finite Element Modelling and Structural Optimization based on Macro-Failure Criterion for Laminated Composite Structures with Free Edge. Int. J. Computational Civil and Structural Engineering. 2000. Vol.1, Issue 1. P. 91104. 34. Borovkov A.I., Palmov V.A., Banichuk N.V., Stein E., et al. Macro-Failure Criterion for the Theory of Laminated Composite Structures with Free Edge Delaminations. Int. J. Computers & Structures. 2000. Vol.76 (1-3). P. 195204. 35. Avdeev I., Borovkov A.I., Kiylo O.L., Lovell M.R., Onipede D. Mixed 2D and Beam Formulation for Modeling of Sandwich Structures. Int. J. Engineering Computations. 2002. Vol. 19 (4). P. 451466. 36. Cescotto S., Charillier R. Frictional Contact Finite Elements Based on Mixed Variational Principles. Int. Journal for Numerical Methods in Engineering. 1992. Vol. 36. P. 16811701. 37. Simo J.C., Laursen T.A. An Augmented Lagrangian Treatment of Contact Problems Involving Friction. Computers and Structures. 1992. Vol. 42. P. 97116. 38. Pisarenko G.S., Yakovlev A.P., Matveev V.V. Handbook about Strength of Materials. Kiev: Naukova dumka, 1988.

- Crystal CathedralUploaded byTheLivingChurchdocs
- Cathedral ArchitectureUploaded byTiwariG16
- The CathedralUploaded byOanaTeodora
- Aeronautical Syllabus-Shivaji University- R.H.B. RamamurthyUploaded byZachary Guerra
- Static & Dynamic analysis of piping systemUploaded bykarthikeashwar
- ejemplo de reporte de flow simulationUploaded byGeorge Ramos Esquivel
- Syllabus for First Year Second SemUploaded bysupriya_apc
- Fluid k2optUploaded bybipin7863674
- Chap 6Uploaded byLarry Cooper
- EFFECT OF VARIABLE VISCOSITY AND THERMAL CONDUCTIVITY OF MICROPOLAR FLUID IN A POROUS CHANNEL IN PRESENCE OF MAGNETIC FIELDUploaded byIJBSS,ISSN:2319-2968
- Fluid Mechanics SyllabusUploaded byRakesh Jha
- Civil Engg. Scheme & SyllabusUploaded byravi007alfa
- Mixed EHL LubricationUploaded byRudrendu Shekhar
- Nonlinear Convection towards an Exponentially Shrinking Sheet with Magnetic FieldUploaded byijsret
- Fluid DynamicsUploaded byYaj Bhattacharya
- WIND TUNNEL DESIGNS, Ahmed.pdfUploaded byAhmed Sabry
- 1.Flow of FluidsUploaded byJohn P. Bandoquillo
- 13904_FULLTEXTUploaded byLuis Miguel Perez Pertuz
- PDFUploaded byDeepak Sudan
- Reynolds Average Navier-Stokes EquationUploaded byhilmanmuntaha
- Week 6 Heat Transfer LectureUploaded bySarindran Ramayes
- Important Topics For GATE.pdfUploaded byTheo
- water-08-00495Uploaded byVictor Peter
- Influence of Turbulence Closure Models on the VorticalUploaded byEternity Immortality
- Course Syllabus AD IUploaded byYsharath Chandramouli
- Implications of ViscosityUploaded byHamza Ahmed
- Tutorial 1Uploaded byKevin Tan
- FFOUploaded byMohammad Imran
- Friction Factors of Power-law Fluids in Chevron-tyUploaded bynghi
- Dhar Anuj - India 39 s Biggest Cover-up 2012 AnujUploaded byAbhishek Palo

- Defects on & in SlabsUploaded byfoush basha
- 07180 - Traffic Coatings - MSTUploaded bywafikmh4
- Inspections Using NA CodesUploaded bylguevara54
- ASTM A751-01.pdfUploaded byeng_diegorossetto
- Spangenberg Et Al-1999Uploaded byHarold Giovano Velásquez Sánchez
- Krones Vario CleanUploaded bycarherji1473
- 1. Dr. Alan Gent - Strength, Wear and Friction of RubberUploaded byIvan Savić
- Enthalpy change of reaction between Phosphoric Acid (H3PO4) and Sodium Hydroxide (NaOH) using Coffee-cup CalorimetryUploaded byValentin-AngeloUzunov
- Tm 10-8415-236-10 Extended Cold Weather Clothing System Generation 3 (Ecwcs Gen 3) Nov. 2014Uploaded byGLOCK35
- Geotechnical InvestigationUploaded byDeepak Kumar Mallick
- Reviewer - Materials Engineer ExamUploaded bybambagen
- Celicoxib.docxUploaded byDavid Raju Gollapudi
- CSM-RE4040-BE-LUploaded bydalton2003
- tut17Uploaded byamir_karimi
- Hardness Observation SheetUploaded byKalasekar M Swamy
- Water Letter General ChemistryUploaded byraghavan89
- tid-001.pdfUploaded byJorge Kovach Alvarado
- Composite Action of Ferrocement Slabs Under Static and Cyclic Loading-composite Action of Ferrocement Slabs Under Static and Cyclic LoadingUploaded byKarrar Monarch
- 5054_s15_qp_22Uploaded byidyllic20
- Torit Dfe Dust Collector SystemUploaded bybmsengineering
- BR_CAT_ENG_130808Uploaded byPeru Haya
- Experiment 6Uploaded byLa Dolce Aisya Ain
- 04IddarUploaded byDivya Naradasu
- Application of Weighted Total Acceleration Equation on Wavelength CalculationUploaded byIJAERS JOURNAL
- Scanning electron microscopy study.pdfUploaded byJavier Olguin
- solid-CHM103-inter.pdfUploaded bySirawit Tachachaikulsiri
- Design and Selection of Deepwell PumpsUploaded byJorge Mártires
- completedissertationUploaded byHariprasad Madhukarrao Paikrao
- Underground Storage of Energy Carriers Brine and Salt ExtractionUploaded byJorge Athié Huitrón Moreno
- 2nd semester final for webUploaded bylshaw