You are on page 1of 10

16587 Pressurised Systems

9.6 FLANGE DESIGN Flanged joints are used on many pressure vessels and heat exchangers and of these, the majority are designed or selected from flange specification standards such as BS15601 or ANSI B16.52. These do not require much consideration of behaviour of gaskets, bolts and flanges in detail, as these aspects are implicit within these codes. Standards of this type have been derived from a combination of experience, testing and calculation within specific industries and provide a useful combination of economy, safety and standardisation. When standard flanges cannot be used, or are not appropriate to the circumstances, then it becomes necessary to design the joint in detail to match specific requirements. BS 5500 provides methods for analysis and design of a range of special joints, allowing the designer to create a design for the most appropriate joint in a given situation. For a flanged joint to seal, the correct combination of geometry, gaskets, bolts, surface finish and materials must all be specified. This task is made more difficult by the practical variability of some of these elements and it is therefore essential that an effective design system is used which encompasses all aspects of joint behaviour and which is capable of producing reliable sealing within known and clearly defined parameter limits. Flanged Joint Behaviour A typical flanged joint as shown in Figure 1 and consists of five inter-dependent elements; bolts, gasket, flange ring, taper hub and shell. In different joint types ,these elements may change in shape or form but they retain their basic functions and perform in a similar way.

Bolts Flange Hub Gasket Shell

Figure 1. Typical Flanged Joint Before examining the detailed behaviour of each element, it is worth examining their relative and fundamental contributions within any joint. Bolts Bolts are used to enable disassembly of a joint during its service life and allow access for maintenance and inspection. They are also required to hold the joint together under pressure and to pre-stress the gasket sufficiently to enable it to function as a seal.

1 2

British Standards Institution, Steel pipe flanges for the Petroleum Industry BS1560. American National Standards Inst, Pipe flanges and flange fittings, ANSI B16.5.
123

16587 Pressurised Systems

Gaskets A gasket is introduced between the flanges to prevent the contained fluid from leaking. It is usually made from a softer and more flexible medium than the flanges and is thereby capable of adapting to the shape of the flange surfaces, making intimate contact, and reducing potential leakage gaps to a minimum. Sealing can thus be achieved at a lower pre-stress and more economically than would be required with two metal flange faces being brought together without a gasket. Gaskets are also convenient because they are relatively cheap and easy to replace and should require minimal rework when in service. Tightening of the bolts with correct pre-stressing of the gasket is vital to the successful performance of a joint.

Flange Ring In general, flanges have to be introduced on the ends of the shells in order to accommodate the bolts. However, because a continuous ring type gasket cannot be accommodated at the same diameter as the bolts, then the offset between the bolts and the gasket causes the flange to be twisted by tightening the bolts. Moreover, because the bolts cannot be introduced in line with the pressure load ie. at the mid-thickness of the shell wall, then a further bending moment on the flange is introduced. This of course means that the flange now also has the job of providing resistance to, and structurally sustaining, bending moments introduced as a consequence of its own character. Taper Hub and Shell The taper hub form the connection from the flange ring to the shell. There is normally a gradual reduction in diameter from the thick end of the taper hub to the main shell wall thickness. The taper hub provides reinforcement at the back of the flange and acts to support the bending moment generated by the flange ring.

Figure 2a,b Flange/Shell Behaviour under Load


124

16587 Pressurised Systems

The previous figure illustrates typical displacements and stress distributions along the shell wall surface when a joint is subjected to pressure. Figure 2a shows the basic loads which typically occur on a joint. The bolt load during tightening is reacted by the gasket producing a bending moment on the flange, and the internal pressure generates a longitudinal load in the shell wall which is reacted by the loads on the already pre-stressed joint, inducing further bending on the flange. Figure 2b demonstrates a typical form of the distribution of radial displacements. An approximately linear variation of radial displacements can be seen across the flange thickness due to bending and the shell wall can be seen to be pushed outwards by its connection to the flange. It can also be seen that the shell displacements decay down to the value of the displacement caused by pressure alone, within a comparitively small distance from the flange. A typical longitudinal stress distribution is also shown for a flange with and without a hub. Note that the disturbance created by the flange on the primary stress systems is localized to an extent within a certain distance of the back of the flange, in practice about 1.5 rt . It can also be seen that for a flange without a hub, a high stress occurs at the joint with the shell but that the introduction of a relatively small hub can reduce this stress significantly. Thus a taper hub can be used to assist the shell by reducing the high stresses between it and the flange to within acceptable limits with consequent potentially significant economies. It is clear from the foregoing that the common type of flange joint relies on brute force to provide sealing with little or no leverage or elegance. However, even with this apparent indelicate simplicity, this type of joint has been used very successfully for many years, perhaps because of its simplicity. This can also be its undoing because it introduces sensitivities to factors, which are often not obvious, and frequently beyond the direct control of the designer. Fortunately, by following the requirements of design codes such as BS 5500, and having an understanding of the basic elements and their interaction with each another, it is possible to design good joints which will provide long life and reliable service. Joint Interaction To properly understand the contribution of the individual components in a flanged joint it is useful to examine interaction between elements in flanged joints. Consider the joint shown in Figure 3. In this the three essential elements of the joint are represented by simple elastic springs with gasket linear stiffness Kg , bolt linear stiffness Kb and the flange rotational stiffness kf. From these, the force displacement relations for the assumed three elements can be written as, Fg = K g . Ug Fb = Kb . Ub M f = k f . f where Fg, Fb, and Ug , Ub, are the loads and displacements in the gasket and bolts respectively and Mf and f are the flange bending moment and flange rotation. The key to the analysis is the estimation of the stiffnesses of the gasket, bolt and flange respectively. The flange rotational stiffness, kf, is the rotational stiffness of a simple ring, ignoring the stiffness contributions of the hub and the shell. The gasket stiffness is somewhat more
125

16587 Pressurised Systems

complicated since the gasket is more flexible during initial tightening than during unloading due to pressure. In addition, during pre-loading of the gasket, the bolts extend and, as a result, squeeze the gasket inducing a bending moment. This is normally referred to as the Bolt-Up case.

Figure 3. Flange joint Interaction Model When internal pressure is applied, this known as the Operating case, the longitudinal pressure load, Fp, tries to pull the joint apart. It also relieves some of the pressure on the gasket and adds some load to the bolts and causes the contact surfaces to rotate. Unlike the bolting-up load case, the internal pressure case is not statically determinate and requires consideration of the equilibrium, constitutive and compatibility relations in order to find the changes in bolt and gasket loads, Fb and Fg, as functions of the applied internal pressure load. Fb = Fg = (1 2(d c)c. K g / k f ) (1 + K g / K b + 2c 2 . K g / k f ) ( K g / K b + 2cd K g / k f ) (1 + K g / K b + 2c 2. K g / k f )

. Fp . Fp

From a simple viewpoint it would normally be thought that applying internal pressure to a joint would always increase the load on the bolts but from an examination of the above equation, it can be seen that bolt load will decrease with increasing pressure, if the top line is negative! This will occur when, 2c(d c ). K g / k f > 1 Thus when the gasket is relatively stiff and/or the flange is excessively flexible, the bolt load will reduce with increasing pressure. From the above equations, it can be shown that the effect of this is that the gasket load drop will be equal to the sum of the pressure end load and the bolt load drop which will result in the gasket pressure falling rapidly as vessel pressure increases, accelerating the possibility of leakage. This concept is an important one in understanding joint behaviour as it fundamentally indicates the importance of component relative stiffnesses and joint interaction and interdependence.
126

16587 Pressurised Systems

Gaskets and Leakage Leakage will occur when the gasket pressure reduces to some gasket minimum sealing pressure (Pgm) which will increase with vessel pressure. Theoretically a joint will seal providing the gasket pressure remains greater than the vessel internal pressure but in practice it is found that in order to have some margin of safety against leakage, it is necessary to keep the gasket pressure at some factor m above the internal pressure P, i.e. Pgm = m. P The gasket factor m is given for various gaskets in Table 3.8.2(2) of BS 5500 with a value of around 3 being found adequate for most purposes. In BS 5500, gasket behaviour is not explicitly described and is only indirectly incorporated via some simple gasket parameters, y and m which allow necessary joint conditions to be derived for a particular gasket selection. From the interaction analysis it is clear that some of the essential characteristics for any gasket are good flexibility together with a large recovery displacement. However, a variety of types of gaskets are available each of which exhibits their own complex characteristic behaviour. These gaskets have been developed over many years to match specific and general industrial and process requirements. Fortunately, a number of useful principles have been evolved, mainly empirically, using reasonably predictable characteristics which give repeatable performance when used with suitable safety margins.

Figure 4. Typical Gasket Behaviour


127

16587 Pressurised Systems

Figure 4 illustrates some of the more common typical gasket characteristics. On first loading, as the bolts are tightened up, the gasket usually follows a non-linear and nonrecoverable path. During this initial phase (O-A) the gasket is forced to conform to the flange faces, absorbing any lack of flatness and filling the irregularities and asperities as are present on any surface. Seal quality will depend on how effective the gasket is at fulfilling these tasks since they minimise available leak paths for the contained fluid. The point at which the gasket provides the minimum effective seal is known as the gasket seating stress (y) and is given for various gasket types in Table 3.8.3(2) in PD 5500. The seating stresses as given have been obtained from a combination of industrial experience and tests on typical gaskets. For many years these values have been used successfully in the design of joints. The region marked A-B-C is the useful sealing range of the gasket. For an effective seal, the joint should be assembled to some stress value between the gasket seating stress (y) at point A and the crushing limit of the gasket at point C. Although the seating stress is given in the code, the crushing limit is not but it can usually be obtained from gasket manufacturers. When the gasket is compressed beyond its crushing limit, some form of breakdown usually occurs in such a manner that joint sealing is adversely affected. The nature of the breakdown varies according to the type of gasket. If the gasket is tightened to some value between A and C and then the gasket is unloaded (by internal pressure or bolt loosening), it will follow a path something like `B-B'. When the gasket is reloaded it will follow a path close to the decompression line but with some hysteresis. When the loading again reaches point B, the gasket then continues to follow the initial loading curve A-C as though it had never been unloaded. This reasonably predictable performance makes it possible to use the simple joint analysis given earlier with the recovery modulus used to describe the gasket behaviour. At some point during the unloading of the gasket it reaches a point at which it can no longer reliably perform its sealing functions. This minimum gasket sealing stress (or pressure) is dependent on the gasket type and the internal pressure. As explained earlier it is usually calculated from the product of the gasket factor m and the maximum internal pressure in order to ensure that the gasket pressure always exceeds the internal pressure. Bolts PD 5500 requires that sufficient bolting is provided to produce enough load for gasket seating in the Bolting-up Condition and that the bolts be designed to absorb the required Operating condition, sustaining both the hydrostatic end force and maintaining sufficient gasket pressure to ensure a leak free joint. In any bolted flanged joint, the bolts are required to provide a preload that ensures gasket sealing performance and maintains the integrity of the joint under pressure. From the interaction analysis it is apparent that for sealing efficiency in a joint with a narrow faced gasket, the bolts should be as stiff as possible. It is essential that bolts perform satisfactorily under the two main conditions of bolt-up and operation . In addition, the bolts must ensure that the gasket operates adequately, satisfying the following:
128

16587 Pressurised Systems

(1) Adequate bolting capacity is provided to ensure that the gasket can be properly seated. (2) Adequate bolting capacity is provided to ensure that sufficient assembly stress can be generated to cover the sum of the minimum sealing stress and the vessel internal pressure end load. Provision (1) makes sure that the gasket reaches at least point A in Figure 4 by the use of the following formula for the minimum required bolt load for gasket seating. Wm 2 = . b. G. y where b is the effective gasket width and G is the effective gasket sealing diameter. It is essential that Provision (2) makes certain that when the gasket is tightened, it has enough preload to ensure that the minimum sealing stress is left on the gasket when the design pressure is applied to the joint. This is covered by the following formula for the minimum required bolt load for operating conditions, Wm1 = G 2 P / 4 + 2 GbmP For bolt sizing, the minimum required area to satisfy the PD 5500 is the greater of, Am = Wm1 W or m 2 Sb Sa

An actual bolt size has to be chosen which can be fitted geometrically on the flange with appropriate clearances for bolt heads and tightening tools. PD 5500 provides a table of allowable bolt stresses for a variety of commonly used bolting materials, based on the ASME VIII equivalent. In the tightening of a joint, the applied preload will be a function of the actual physical bolt size and not the theoretical minimum required bolt size. If the theoretical minimum were considerably smaller than the actual bolt size and flange stressing was performed and checked with this, then the flange could be significantly overstressed when the actual bolt size preload was applied. To limit this, flange calculations use the average of the actual bolt area and minimum required bolt area (a compromise between economy and safety) for the bolting-up load W, ie. W = ( Am + Ab ) . Sa 2

129

16587 Pressurised Systems

Flange Analysis Calculation of suitable dimensions for the flanges to PD 5500 is performed in a similar manner to ASME VIII using a slightly adapted form of the Taylor Forge method developed by Waters et al.3 published in 1937. The analysis is explained in some detail by Singh4 and will not be repeated here. Some of the principal assumptions and simplifications involved in this method may be summarised as follows, 1. Materials of all of the elements are assumed to be homogenous and remain elastic under the loading conditions assumed in the design. The effect of the bolt holes in the flanges is neglected. Axial symmetry is used to reduce the problem to consideration of the conditions on a single flange, hub and shell cross section, neglecting variations due to location of bolts. All loading applied to the flange is reduced to a couple involving a pair of equilibrating loads located at the extremities of the flange. Stretching of the middle surface of the flange ring due to the applied couple is negligible. Displacements of the joint are small such that the theorems of superposition are valid. When a ring moment is applied to the flange, the point of connection between the flange and the hub is assumed to have zero radial displacement. Hub and shell are assumed to act as thin shells. The inside bore of the hub and shell is used in the shell theory analysis instead of the mean thickness diameter.

2. 3.

4.

5. 6. 7.

8. 9.

10. Effects due to interaction of elements are neglected. The first part of the analysis involves the flange. This is modelled as an annular plate of outside diameter A, with a central hole of diameter equal to the vessel bore B. Loading is applied as a pair of equal and opposite line loads located at the flange extremities as shown in Figure 5. This method of analysis makes the solution relatively simple but it requires the actual gasket and bolt loads to be converted into equivalent loads. This is taken care of in PD5500 by the equations for the flange bending moments which are then implicitly converted in the calculations. For most flanges where the face width is of the same order as the flange thickness this assumption should have little effect. A simple ring analysis would probably have been good enough for many pressure vessels, but the more complicated plate theory was included to cover thinner and wider low pressure flanges commonly used on pipework.
3

Waters E O; Wesstrom D B; Rossheim D B and Williams F S G: Formulas for Stresses in Bolted Flanged Connections, Trans ASME, Vol 59, 1937, p 161 4 Singh K P and Soler A I: Mechanical Design of Heat Exchangers and Pressure Vessel Components, Arcturus Publishers, New Jersey, 1984
130

16587 Pressurised Systems

The equations which result from the plate theory are rather complex but generally only involve one parameter K, K = A/B which can be found as the sole parameter in the equations for T, U, Y and Z in PD 5500.

Figure 5. Equivalent Flange Loading In order to make the design simple, Waters et al3 produced a set of curves similar to those in PD 5500 which made it possible to determine the highest flange stresses without repeating all of the above steps. The method, although iterative, makes possible the calculation of the necessary flange thickness from the known parameters. It is interesting to note that virtually all optimum joint dimensions are set by the geometric constraints for a particular vessel and that the flange thickness is more or less the only dimension that needs to be varied to meet the necessary flange stress limits. The hub stress calculated in PD 5500 and ASME VIII is the hub bending stress due to the applied flange moment. The application of internal pressure generates a longitudinal direct stress which is not included explicitly in the above hub bending stress. However, some cognisance is taken of internal pressure in the equations which are used to derive the flange bending moments. In PD 5500, loading is applied in the theory using the bending moments from the operating and bolting-up conditions. The total flange moment under operating conditions is, M OP = H D .hD + H T .hT + H G .hG Each of the load components can be found using the above figure. HG is the gasket minimum sealing load as given in the second part of the equation for Wm1 which is considered to be located at the gasket effective diameter i.e. at a distance hG from the PCD. HD represents the pressure end load on the vessel due to pressure acting across the vessel bore. This is taken as acting at the centre of the large end of the hub. This may appear somewhat inconsistent considering that the shell model centreline was assumed to act at the
131

16587 Pressurised Systems

flange bore, however in reality the pressure end load is likely to be transferred through the centre of the hub to the flange and thus should give something nearer the correct flange moment. HT is the effect of pressure acting on the annulus of the flange face between the bore and the effective gasket diameter. This force is assumed to act at the point midway between the flange bore and the gasket effective diameter. Bolting-up conditions are covered by considering the load acting as a couple between the gasket and the bolts i.e. Matm = W hG Previous versions of PD5500 used the largest of the two bending moments for the design after correcting the operating moment to an equivalent ambient temperature moment, however in the 1991 version of PD 5500, both operating and bolting-up moments are to be considered separately and the results compared with their respective stress limits in a similar manner to ASME. Since the calculations are in a complex form, working sheets are provided in PD 5500. In PD 5500 three stresses are taken from the analysis in order to determine flange acceptability for the operating and bolting-up cases separately. These are the longitudinal hub stress SH, radial flange stress SR and the tangential flange stress ST, which are limited by, S H <1.5 or S R and S HO or or 1.5 S FO

S H <1.5 S HA

1.5 S FA or S FA or S FA or S FA

0.5 (S H + S R )< S FO

S T < S FO

0.5( S H + S T )< S FO

where, SFO, SFA, SHO and SHA are the allowable design stresses for the flange and hub materials, at the operating and ambient temperatures, respectively. From these it can be seen that since the allowable design stress is usually about 2/3 of the material yield, then this allows the hub to be stressed up to the material yield point, allowing yielding in the hub during hydrotest. The flange stress limits are set to a level which should keep the main flange bodies elastic under all conditions, providing the joint is not overtightened during bolting-up. The latter two stress limits are the application of a Tresca type criterion to the bi-directional stresses at the interface between the flange and hub.

132

You might also like