Britt Engineering Associates, Inc. Birmingham. Alabama www.beacom.com September 30, 2018 Technical Note Subject: FRP Flange Design Ref: Thomas E. Graham, FRP Flanges for Process Pipe and Tanks, NACE, 1989 Background In 1986-87 we encountered some major flange failures that developed after about a year of service in a new bleach plant in NC. We performed the pipe stress analysis and provided assistance to the client’s purchasing department during the bidding process. The low bidder offered the two piece flange even though integral flanges were specified the bid was accepted by the client. Also the client had experienced earlier problems with filament wound pipe and designated contact molded for all systems. Some of these flange problems are presented in the NACE paper along with incidences occurring at other plants that we were asked to analyze. Graham’s paper was received with great interest and more than 100 copies were distributed to the attendees of the 1989 conference. A US University and a German University asked permission to use the information in a mechanical engineering class. NACE was contacted and gave approval to use the copyright material. Tom passed away some years ago but he conducted a number of seminars that were requested by several engineering firms. I believe the information is relevant today and may be of interest to engineers who are engaged in FRP piping system design. Flanges are one of the weak links in piping. Sincerely, Frank Britt PE Attachments: 1. Thomas E. Graham, FRP Flanges for Process Pipe and Tanks, NACE, 1989 2. ADDENDUM
26
Embed
Flange Graham MEMO - beacomp.combeacomp.com/wp-content/uploads/2018/01/Flange-Paper.pdf · efforts to use the Taylor-Forge flange design procedure for flat face flanges (3) normally
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Britt Engineering Associates, Inc. Birmingham. Alabama www.beacom.com September 30, 2018 Technical Note Subject: FRP Flange Design Ref: Thomas E. Graham, FRP Flanges for Process Pipe and Tanks, NACE, 1989 Background In 1986-87 we encountered some major flange failures that developed after about a year of service in a new bleach plant in NC. We performed the pipe stress analysis and provided assistance to the client’s purchasing department during the bidding process. The low bidder offered the two piece flange even though integral flanges were specified the bid was accepted by the client. Also the client had experienced earlier problems with filament wound pipe and designated contact molded for all systems. Some of these flange problems are presented in the NACE paper along with incidences occurring at other plants that we were asked to analyze. Graham’s paper was received with great interest and more than 100 copies were distributed to the attendees of the 1989 conference. A US University and a German University asked permission to use the information in a mechanical engineering class. NACE was contacted and gave approval to use the copyright material. Tom passed away some years ago but he conducted a number of seminars that were requested by several engineering firms. I believe the information is relevant today and may be of interest to engineers who are engaged in FRP piping system design. Flanges are one of the weak links in piping. Sincerely,
Frank Britt PE Attachments:
1. Thomas E. Graham, FRP Flanges for Process Pipe and Tanks, NACE, 1989 2. ADDENDUM
FRP FLANGES for PROCESS PIPE and TANKS
By
THOMAS E. GRAHAM PE
Consult ing Engineer
Presented at
NACE
Tenth Biennial
MANAGING CORROSION WITH PLASTICS
SYMPOSIUM
November 6 1989
San Antonio, Texas
I . A B S T R A C T :
This paper discusses details of manufacture, design, and application of contact molded flanges used in process piping and tanks and includes case histories and stress analysis of typical installations. Several failures will be discussed along with an analysis of the causes. Fabrication methods that have improved reliability and strength are presented.
I I . I N T R O D U C T I O N :
FRP f langes become necessary when joining pipe to tanks, pumps, valves and other equipment. This necess itates that FRP f langes match metallic f lange standards in regards to bolting patterns and f lange diameters. Due to the lower modulus of elasticity of FRP, along with other unique characterist ics of f iberglass, this requires careful design and good fabrication techniques in order to achieve a seal under all service and environmental conditions. This paper deals with the requirements for FRP f langes and some designs that have worked and discusses some designs that have given problems in the past. Only contact molded f langes are considered. The main object ive of a f langed joint is to affect a seal between the f iberglass pipe and the component to which it interfaces. To prevent leakage in a gasketed joint of any design, a bolt force is required to compress the softer gasket material so that the gasket seating surface fills the asperities of the contact faces of the two mating surfaces. Leakage is prevented only if the gasket material actually f ills all depressions of the seal contact faces and the f lange-hub assembly is suff iciently rigid that it is not distorted under the bending and torsional effects of the eccentric loading of the f lange due to pressure, gravity, temperature, and bolt pre-load. This seal must be maintained throughout the life of the joint. Care must be exercised to prevent "blow-out" of the gasket due to low sealing stress, especially on flanges that have a very slick surface such as those la id up on a g lass surface . The same results can occur if low durometer gaskets are used for high pressure service. Gaskets of 40 durometer should not be used in pressure service above 50 psi. (4)
I I I . F R P F L A N G E R E Q U I R E M E N T S :
A. PS-15-69: When an engineer needs to design a f lange he has a limited number of resources to work with. NBS VOLUNTARY PRODUCT STANDARD PS-15-69 has been, and continues to be, the main source for f lange information for FRP process pipe and duct. But, PS- 15-69 begins in Section 3.5.7. on f langes, with the statement "The use of f langes shall normally be kept to a minimum..." this doesn ’t give the engineer a very confident feeling about designing f langes. Then he looks in Table 5 for the thickness requirements and f inds that the table is not f i l led out for large pipe diameters and high design pressures.(1)
B. ASME SECTION VIII: Having run into this dilemma he turns to the ASME CODE SECTION VIII for BOILER AND PRESSURE VESSEL DESIGN only to find the following under ARTICLE 3-4 entitled FLANGES WITH OTHER THAN RING TYPE GASKETS, "The rules in Article 3-3 shall not be construed to prohibit the use of other types of bolted f langed connections, such as f langes using full-face gaskets or other means of f ixing or clamping the
f lange at the bolt circle to provide effective restraint against f lange deflection. Such designs may be used provided they are designed in accordance with good engineering practice and the method of design is acceptable to the Inspector." [2]
C. Taylor-Forge method of design: Fortunately there have been a number of efforts to use the Taylor-Forge flange design procedure for f lat face f langes (3) normally used in FRP piping systems. The background for this method is in metallic flanges. FRP is an anisotropic material whereas the theoretical and testing work or f langes is based on orthotropic material. The question has to be raised as to whether the procedure is directly applicable. FRP flanges that have been designed and used in the past become a bench mark in evaluating the design methods for sizes and service condit ions within this experience range. However, much needs to be done in the area of f lange test ing and analyt ical work to bring f langes within the realm of component design that is based on sound design standards that assures the engineer that his design and specif ication of f langes wil l result in a reliable system instead of a weak link.
IV. FRP FLANGE DESIGN:
TAYLOR-FORGE METHOD: Let us look at a typical f lange to see what potential problems might arise and what variables the designer has to work with that can result in a reasonably sound design that can be fabricated with as little difficulty as possible. Figure 2 shows a f lange that is typical of that used for FRP pipe. Since the bolting pattern is set by standards for metallic pipe that usually mates to the FRP flange and the f lange OD is set within certain limits, that leaves the f lange thickness and the hub reinforcing as variables that can be changed to meet the design requirements. There does no t appear to be a restrict ive limit on f lange thickness except economy in design, but there is a l imit on the hub reinforcement thickness because of the bolt location and a possible interference with the installation of bolts. (See the right side of Fig. 1.)
The stresses that a f lange undergoes are illustrated in f igure 1 (see Appendix). There are four primary stresses that normally determine the design of the flange. These are, (1) SH, which is the longitudinal stress in the pipe wall and hub area, (2) SR which is the radial stress in the f lange at the inside diameter of the f lange, (3) ST which is the tangential or hoop stress acting on the cross sectional area of the flange and is maximum at the inside diameter of the flange acting in conjunction with SR, and (4) SR A D which is the radial stress at the section where the bolt holes remove material and weaken the section. (See the Appendix for nomenclature used in the analysis.) These stresses result from installation and operating conditions which includes loads from internal pressure, thermal effects, beam bending act ion, and preload from the bolts. Some of these loads are direct loads that cause axial tensile stress and some cause bending due to being eccentrically applied and result in bending stresses that can add to the axial tensile stress. For instance, all tensile loads that are transmitted through the flange to the mating flange must travel through the bolts since this is the only structural system that is capable of res ist ing these tensi le loads. The loads are t ransmitted back to the f lange skirt which results in bending stresses in the f lange at the bolt locations. See Fig. 2 for flange loading assumptions for stress analysis.
Now look at a typical design that i l lustrates a procedure for controlling stresses within limits that can be tolerated for f iberglass pipe. PS-15-69 has a
foot note under TABLE 5 for f lange thickness select ion which states that, “This Table is based on a factor cf safety cf 8 TO 1 and a flexural strength of 20,000 psi. This latter value is slightly under the minimum f lexural strength for laminates of 3/8 inch and up (see Table 1), due to the manufacturing technique."[1] Other requirements of PS-15-69 include the hub reinforcing length to be equal to four (4) times the f lange thickness and the hub reinforced thickness equal to 1/2 the f lange thickness as a minimum. The example problem will be performed on a 42 inch f lange at 150 psi to i l lustrate the problems that occur in designing large diameter, high pressure flanges.
The matter of hub reinforcement thickness is subject to some interpretat ion. Some have assumed that the requirement is that the tapered portion of the hub outside of the structural wall of the pipe be equal to 1/2 the f lange thickness as indicated in ASTM Standard D 3299 for tanks, (5) Table 4 of this Standard shows this as a requirement for contact-molded flanged nozzles. However, the table only applies to 25 psi rating. If the same requirements are applied to 150 psi f langes, then the hub reinforcement extends out into the bolt hole area and much of it is cut out when spot-facing for the washers and bolts. This cutout material is not accounted for in the design procedure and thus the f lange is weakened to a point below the strength ind ica ted by the design procedure. PS-15-69 does no t specifically state that this is a requirement; but, when flanges are molded onto a piece of pipe, it becomes a requirement due to the nature of the fabrication. This type of f lange has had problems in the past on some projects due to its two piece construction and the weakened area at the secondary bond interface. Our example problem is for an integral hub f lange with one piece construction.
Three cases are calculated based on the hub reinforcement requirements, (1) The hub dimension equal to 1/2 the flange thickness plus the pipe wall thickness, (2) the hub thickness equal to 1/2 the flange thickness, and (3) the hub thickness set to clear the spot face dimension as much as possib le. The results are summarized in the following table:
CASE 1 CASE 2 CASE 3
FLANGE THICKNESS MAX HUB THICKNESS INTEFERENCE BETWEEN HUB AND SPDTFACE MAX. STRESS
I t is ev ident when look ing at these th ree cases tha t the hub thickness is critical in limiting the longitudinal stress at the hub-f lange intersection. The f lange thickness has to increase much more than the reduction in the hub thickness as shown in the three cases. The wall thickness for 150 psi rated HLU pipe necessitates that the spot-face at least get into the fillet area of the hub-flange junction, even when no hub is used as for case 3.
FRP FLANGES for PROCESS PIPE and TANKS
6
V. DESIGNS THAT HAVE HAD PROBLEMS:
A. Twenty inch stock line in bleach plant: The flanges involved were for 12 inch, 18" and 20" pipe. The project pipe fabrication was split into two orders to expedite pipe shipment. Part of the pipe was hand-lay-up and part was f i lament wound. A l l of the f langes were hand-lay-up with stub flanges except in several cases where pipe routing dictated that flanges be made onto some 90 degree elbows in order to f it in the cramped space (see Fig. 5.) The f langes were photographed and identif ied as to location.
The f lange construction for the pipe is shown in f igures 3 and 4. One Fabricator laid the f lange up directly on a straight pipe, sect ion with the hub reinforcement thickness equal to one half the f lange thickness when measured from the OD of the pipe (see Fig. 3). The pipe for this f lange was 150 psi contact molded pipe and the f lange thickness was 2 5/16 inches from face of f lange to the spot-face surface. The spot-facing for the bolt holes cut into the hub area as shown in Figure 3b. The other fabricator laid the f langes upon tapered pipe or f ittings as shown Figs. 4 and 5. An effort was made to minimize the hub thickness to avoid interference with the spot facing.
There were numerous cracks at the centerline of the bolt holes in both directions as well as at the hub-f lange intersect ion. The major cracks that led to leakage were in the f langes that were fabricated according to the technique shown in Fig 4, whereas the cracks in the flanges constructed according to Fig. 3 appeared to be superficial even though they were still of a concern. Much of this piping has since been replaced due to process up grades.
A major failure occurred in a 20 inch flange that was made directly on a 90 degree elbow using the technique shown in Fig. 4 and 5. The line was a chlorinated stock l ine operating at 87 psi and 160 degrees F. It was located at the bottom of a 90 feet high chlorination tower and was a critical line, both from a process and safety view point. Fortunately, the failure was detected when it first began to leak badly and temporary reinforcement was added to prevent a catastrophic failure until a new elbow section could be fabricated using the technique shown in Fig. 3 and installed. Other contributing factors to the cracking and failure of the flanges were the thin, hard gasket material used along with the high bolt torques that were required to affect a seal. Flat face flanges with full face gaskets require a sealing stress ever the entire area of the gasket which results in a large bolt load if the gasket has a high durometer. A low durometer gasket should be used to reduce the seating stress of the gasket and thus result in lower bolt pre-load to effect and maintain a seal.
B. Twenty four inch diameter water line:
This case involved a 24 inch f lange that failed due to bending stresses that resulted f rom unrestrained expansion joint that induced bending from pressure acting on a short offset. This f lange was made up on a stub f lange section without any hub reinforcement (see f ig. 6). The system was designed for 25 psi, but actually operated at about 35 psi. The flange failed upon start-up because there was an unrestrained
expansion joint at the pump flange which isolated the FRP pipe f rom pump vibrat ions. The yield ing of the expansion joint under pressure created a high bending moment on the flange and it cracked immediately. It was replaced with a higher rated f lange and restraints were added to the expansion joint.
C. A sixty-six inch diameter cooling water l ine for a power plant:
This f lange was 6 inches thick and built on a piece of pipe with two piece construction. It developed cracking at the f lange-hub junction when the restraining bolts at an expansion joint were left loose and the pressure caused the elbow to move and induced a high bending moment. Fortunately the f lange was supported with a component support that acted as a semi-anchor and reduced the bending moment on the flange.
D. A forty-two inch flange for a 100 psi salt water pumping station:
These f langes were fabricated 4 inches thick using one piece construction. The Contractor called to complain about losing the gasket due to blow-out even though he had used the recommended torque supplied by the fabricator. The faces of the flange had been made on a glass table and were very smooth and slick. This resulted in blow-out of the gaskets even though the bolt torque was adequate for ordinary 100 psi service. After some calculations, it was decided to increase the bolt torque 50 foot pounds f rom the original 120 foot pounds specified. All the joints sealed except one. It was taken apart and some debris was found between the flanges that caused the leakage. The f langes sealed with the addit ional bolt loads without failure.
VI. RECOMMENDED FLANGE DESIGN AND DETAIL:
Fig. 7 shows a method of f lange construct ion that the E n g i n e e r i n g d e p a r t m e n t h a s r e c o m m e n d e d a n d u s e d o n p r o j e c t s f o r m a n y years. This flange is a one piece, integral unit. The hub reinforcement is built up to a thickness equal to one half the f lange thickness. Woven roving is made continuous from the shell and hub into the flange as well as the mat layers. The additional thickness of the flange is made up using donut shape sections of mat alternated between the layers of woven roving.
Table 1 presents dimensions for this type of f lange for sizes 2" through 42" and pressure ratings from 25 psi through 150 psi. In the calculations an attempt is made to minimize the interference between the spot-facing for the bolt holes and the hub reinforcement by setting the hub reinforcement to clear the spot-face diameter and determining the f lange thickness required to bring the stresses ti: 2500 psi or less. A minimum hub shell thickness is set at 1/2 the f lange thickness. Thus, for larger diameters and pressures, the hub stil l encroaches into the spot-face area for the bolts but is held to a minimum.
It can be noted that the f lange thicknesses are greater than those listed in Table 5 of PS-15-69 for many flanges in the table. If the initial assumption had been based on using predetermined f lange thicknesses and determining the hub thickness required the flange thickness would have been the same as in PS-15 where shown. Thus, it becomes obvious when performing analysis on FRP f langes that the geometric restraints for mating up to steel flanges dimensionally creates many prob lems. Th is is compounded by the fact tha t FRP analysis is being performed by procedures and with coefficients that were derived through many years of research, testing and design of steel f langes. The same test ing and research needs to be performed on fiberglass flanges to determine the proper procedures, proportions and factors of
safety needed to provide reliable f langes. Such test ing would need to be performed on full scale test samples under realist ic condit ions as they occur for FRP pipe installations. This would involve all sizes and pressure ratings acting under the combined effects of pressure and external bending on the f langes at design temperatures. Various gasket materials should be tested to determine the optimum durometer gasket to use and the optimum bolt torque to affect a maintainable seal.
Table 2 gives the results of calculated bolt torques as determined by the analysis that Table 1 is based on. The calculated torques are based on the following:
Maximum calculated bolt torque: Bolt load = W = .5 (Am+ Ab) Sa Bolt torque = (k P Bd] 1/12, where: Bolt torque is in ft. -lbs.
k = .15 for lubricated bolts. Testing indicates that this value can vary between .10 and .23
It should be noted that bolt torques for pressure piping flanges are affected by many factors, such as type of gasket, length and type of bolts, f lange elast ic modulus and rig idity, temperature, lubricity of bolts, etc. All gaskets creep after they have been loaded by the flanges which results in reduced bolt load. Creep is the tendency for the material to continue plastic deformation with no increase in load. This creep relaxation is greater at higher temperatures. Since most piping systems operate with a f luid temperature that is higher than the ambient temperature at installation, creep due to increased temperature is of concern. It has been reported that 90% of creep takes place in the f irst 24 hours at operating temperature. This makes inspection after start-up an important tool to assure proper bolt torques and other installation parameters have been performed properly. If leaks do occur, the system pressure should be relieved prior to re-torquing the bolts. Re-torquing should be done in approximately 10 f t-lbs increments using the cross-torque method until the leak is stopped. If the maximum bolt torque value in table 2 is reached without stopping the leak, the engineer needs to be consulted to determine the cause to avoid over stressing the flange.
Poorly designed, fabricated or installed FRP f langes are a weak link in a piping system. Attention must be paid to many aspects of design, fabrication, and installation to be assured of a reliable f lange. Short cuts or untried methods can lead to trouble. It does not pay to cut corners on f langes to save a few dollars if the piping system is critical to plant operations. Among the things that must be considered to obtain a good installation are,
(1) Select conservative proportions for the f langes such as those in table 5 of PS-15-69. Increase the thickness if the flange is to be mounted on a thin wall pipe such as f i lament wound pipe that has a structural wall below that listed in Table 3 of PS-15-69.
(2) Use one piece fabrication with as much integral hub reinforcement as practical on stub f langes. Avoid building flanges on pipe sections and f ittings if possible.
(3) Select a soft gasket with a Shore A or Shore A2 hardness between 40 and 70.
(4) Use the proper bolt torque to seat the gasket without overloading the f lange. Place a washer under the head and nut of the bolts that are in contact with fiberglass to avoid a bearing stress under the washer that exceeds 2500 PSI.
(5) Provide good support close to the flange if it acts as a beam to avoid high bending moments on the joint. Supports located at the inf lection point (point of zero moment would be ideal). Finally,
(6) use good installation procedures such as keeping the sealing surfaces clean, don't allow the gasket to become twisted or crimped, lubricating the bolts, t ighten the bolts using the cross torque method and incrementally tightening the bolts until the proper torque is obtained, and inspect the flanges after installation, hydro testing, and start-up to assure that there is no visible damage or leaking.
Good flanges begin with good design and specification and are assured through continuous inspection throughout the fabrication and installation. The extra cost of a quality f lange will provide peace of mind to the engineer and plant maintenance superintendent.
The following symbols are used in the equations for the design of flat-faced flanges employing full-face gaskets.
A = outside diameter of flange, in. (mm) AB = total cross‐sectional area of bolts at root diameter of thread or section of least
diameter under stress, in. (mm)
Am = total required cross‐sectional area of bolts, the greater of Wm1/Sb or Wm2 /Sa, in.2
(mm2)
B = inside diameter of flange, in. (mm) b = effective gasket width or joint‐contact‐surface seating width, in. (mm) C = diameter of bolt circle, in. (mm) d = shape factor for integral type flanges
=
d1 = bolt hole diameter
e = shape factor = F/ho
F = shape factor (Figure M1-300.3.3) f = hub stress correction factor (Figure M1-300.3.4)
= 1 for calculated values less than 1 G = diameter of gasket load reaction g0 = thickness of hub at small end g1 = thickness of hub at back of flange H = hydrostatic end force HD = hydrostatic end force on area inside of flange HG = difference between bolt load and hydrostatic end force HGy = bolt load for gasket yielding
= bπGy H′Gy = compression load required to seat gasket outside G diameter Hp = total joint‐contact‐surface compression load
= 2bπGmp Hp′ = total adjusted joint‐contact‐surface compression for full‐face gasketed flange, lb
=
HT = difference between total hydrostatic end force and the hydrostatic end force area inside of flange
= H − HD h = length of hub, in. hD = radial distance from bolt circle to circle on which HD acts hG = radial distance from bolt circle to circle on which HG acts
= radial distance from bolt circle to gasket load reaction
= flange lever arm
hT = radial distance from bolt circle to circle on which HT acts K = ratio of inside flange diameter to outside flange diameter
L = length of flange including hub M = unit load, operating, lb
= Mmax/B Ma = moment under bolt‐up conditions MD = component of moment due to HD MG = component of moment due to HG Mo = total moment MT = component of moment due to HT m = gasket factor N = number of bolts p = design pressure Sa = allowable bolt stress at ambient temperature Sb = allowable bolt stress at design temperature SFa = allowable flange stress at ambient temperature
SFo = allowable flange stress at design temperature SH = longitudinal hub stress SR = radial flange stress SRAD = radial stress at bolt circle ST = tangential flange stress T = shape factor (Figure M1-300.3.5) t = flange thickness tn = pipe wall thickness U = shape factor (Figure M1-300.3.5) V = shape factor (Figure M1-300.3.2) Wa = flange design bolt load Wm1 = minimum bolt loading for design conditions Wm2 = minimum bolt loading for bolt‐up conditions Y = shape factor (Figure M1-300.3.5) y = gasket unit seating load Z = shape factor (Figure M1-300.3.5)