Hydraulic Tensioner Assembly: Load Loss Factors and Target Stress Limits

Proceedings of the ASME 214 Pressure Vessels & Piping Conference PVP214 July 2-24, 214, Anaheim, California, USA PVP214-28685 Hydraulic Tensioner Assembly: Load Loss Factors and Target Stress Limits Warren Brown Integrity Engineering Solutions Dunsborough, Western Australia wbrown@integrityes.com ABSTRACT This paper examines two key themes regarding the use of hydraulic tensioners to assemble pressure boundary bolted joints. The first theme is the amount of overload required to compensate for load loss that occurs due to both nut-totensioner mechanical interaction and, in the case of less than 1% tensioning, bolt-to-bolt mechanical interaction during assembly using hydraulic tensioners. The second theme is to examine the effect on the required assembly procedure if the target bolt stress value is close to yield. This paper has practical application for general assembly of large diameter bolts, but in particular for assembly of low strength bolting, such as A193-B8M class 1 and class 2 bolts. INTRODUCTION The use of hydraulic tensioners to tighten pressure boundary bolted joints is relatively common practice, particularly when large diameter bolts are involved. There is a general perception in industry (for example EN1591-1 [1]) that hydraulic tensioners are significantly more accurate than other methods, such as torque tightening. However, like any assembly technique the final achieved accuracy is not only a factor of the method being used but also of the entire joint assembly procedure, including number of passes performed. Even the use of 1% coverage of hydraulic tensioners involves a certain amount of guess-work on the part of the operator in order to determine the final residual bolt load once the load has been transferred from the tensioner to the nut. This is because as the load is transferred to the nut, the threads deflect, which releases some of the load originally established by the tensioner. This loss in load is commonly termed the load loss factor or tensioner load loss factor and is compensated for by applying a higher load to the tensioner than the desired residual bolt load, such that once the load loss has occurred, the residual bolt load is approaching the desired target bolt stress. In this paper we will refer to this load loss factor as the Nut Load Loss Factor (NLLF). The problem is further compounded if, as is common, the joint is being assembled using only 5% tensioner coverage or less. 5% tensioner coverage is where half the bolts are tensioned simultaneously and then the tensioners are switched to the second set of bolts and those are subsequently tensioned. In most cases, the tensioning procedure will call for Pressure A, which corresponds to the load applied to the first half of the bolts, to be higher than Pressure B, which corresponds to the load applied to the second half of the bolts. Both of the pressures correspond to loads that are higher than the desired residual bolt load. The first half of the bolts is tensioned higher than the second half in order to compensate for a second source of load loss; the bolt-to-bolt load loss factor, termed the Bolt Load Loss Factor (BLLF) in this paper. This occurs due to mechanical interaction across the joint as the second set of bolts is tightened. The load on the first set drops off due to additional deflection of the flanges and gasket as the second set is tightened. The problem becomes even more complex when fewer tensioners than 5% coverage are used. In fact, as the number of tensioners used is reduced, the accuracy of the method decreases substantially (due to the additional interaction occurring) and/or the length of the required procedure to achieve the same accuracy increases substantially, to the point where torque assembly may become equally as accurate and often substantially faster. 1 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

Given the criticality of pressure boundary bolted joint assembly (particularly joints incorporating larger bolts, which are those most commonly tensioned) and that hydraulic tensioning is not a new technique, one might expect that the two load loss factors would be relatively well understood. However, this is not the case and although there exists some guidance on the matter from tensioner manufacturers (for example, Tentec [2] and SPX [3]), it does not take long to identify that, since the flange rigidity is not considered in any of the guidance, there has been little effort published that is aimed at accurately determining the load loss factors across the range of applications commonly seen in industry. This then plays into the second theme of this paper, which is to examine whether tensioning to near yield values, particularly for B8/B8M bolts, can be achieved accurately. This topic has recently been the subject of some discussion within industry (for example, Noble [4]). It relates directly to the joint integrity in that if it is not possible to assemble the joint to a high enough bolt load, there will be risk of joint leakage. The two topics are linked since if the NLLF is significant, then in order to achieve a near-yield residual bolt load, the initial tensioner load may need to exceed the Specified Minimum Yield Stress () of the bolt. Exceeding the may not be something that, on first consideration, everyone would be comfortable doing, since it raises concerns regarding nut or bolt failure or excessive deformation of the threads. However, with due consideration of the risks and limitations, it is something that can, in many cases, be done with no associated risk. To examine these themes further, the discussion will be broken into two sections, the initial being 1) Can I assemble joints to near residual bolt load levels and the second being 2) Do I want to assemble to near levels. In other words, we will look at the physical limitations on tensioners when assembling to near levels initially and, assuming it is possible for at least a certain class of fastener, we will then look at other possible consequences associated with assembling to near. QUANTIFYING THE LOAD LOSS FACTORS The concept of mechanical interaction in bolted joints has been studied extensively and many papers have been published on the subject. However, none have looked at adapting the approach to hydraulic tensioner joint assembly. However the same concepts, as outlined in Brown [5] for example, can be easily adapted to tensioner assembly. For example, the NLLF occurs due to the deformation of the nut as the load is transferred from the tensioner to the nut. If the system of joint, nut and bolt is represented as a series of springs (Fig.1), then by summation of both the forces and deformations between the two states of the system (tension applied and then residual remaining), the remaining load can be determined as a function of the relative rigidities of the joint components: BL R = R 1. BL I [1] 2h G 2 q f +q b +q g where R 1 = 2h 2 G q f +q b +q g +q N n b R 1 = Nut Load Loss Factor (NLLF) BL R = residual bolt load (N) BL I = initial bolt load (N) h G = flange moment arm (mm) q f = the flange rigidity (rad/n) q b = bolt rigidity (mm/n) q G = gasket rigidity (mm/n) n b= number of bolts and q N = nut rigidity (mm/n) The NLLF is then be expressed by the term (1-R 1). This then allows a more accurate determination of the NLLF based not only on the length to diameter ratio of the bolt, but also on the flexibility of the joint in question. Each of the component rigidities have been previously defined (Brown [5]), with the exception of the nut rigidity (q N). This must be determined by experimentation, as outlined later in this paper. The BLLF can be determined for the 5% tensioner coverage case (or other coverage cases) by using the same spring series representation and by summation of the forces and moments. A similar relationship between the initially applied loads and the residual loads can be determined: BL R,A = R 1. BL I,A R 1. R 2. BL I,B [2] where R 2 = h G 2 q f +q g h G 2 q f +q g +q b R 2 = Bolt Load Loss Factor (BLLF) BL R,A = residual bolt load, pass A bolts BL I,A = initial bolt load, pass A bolts BL I,B = initial bolt load, pass B bolts So, by using the above equations, it is possible to determine the residual bolt stress for both the pass A and pass B bolts after the second pass of the tensioning procedure, based on the bolt stress applied by the tensioner on each pass. By similar methods, it is then possible to continue further and determine the residual bolt load for the bolts after the third, fourth and subsequent passes. Likewise, it is possible to rearrange the relationships to determine the exact tensioner over-load that must be applied for each pass of tensioning in order to obtain the desired final bolt stress. However, before being able to use the method, it is necessary to establish a value for the nut rigidity (q N) by experimentation. 2 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

DETERMINING NUT RIGIDITY The nut rigidity is relatively easily determined by tests using a load cell, long stud bolt and solid steel cylinders (Fig. 2). The test procedure consists of using a tensioner to tighten the bolt in load increments and at different steel cylinder lengths in order to build a matrix of results relating to the different component stiffness for that particular bolt size. This was done for several bolt sizes, materials and load magnitudes. Then using Eq. 1, and the calculated cylinder and bolt rigidities, the nut rigidity was isolated in each case. An example of the test results for a B8M class 1 bolt is shown in Fig. 3. It can be seen that the NLLF increases at low bolt stress levels, but is reasonably stable at normal assembly bolt stress levels. Of interest for the latter part of this paper is that it also maintains similar levels from around 2 MPa to well above and, in fact, all the way past actual material yield. Similar test results from an identical test are shown in Fig. 4. The reason for presenting these results are to once again illustrate the effect of bolt stress on the variance of NLLF, but also to demonstrate that there is inherent variance in the NLLF itself, with the value ranging between.7 to.8 in the stress range of interest. This is the natural variation in the test results and will also be apparent in the field as well as residual inaccuracy in the final achieved bolt stress. Test results for B8M class 2 bolts are presented in Fig. 5 and show a similar trend, indicating that there does not appear to be any material based variance in the NLLF. The final point to note from these tests is that the NLLF was generally fairly consistently found to be around 3% to 8% higher for the first cycle of tensioner pressure and nut tightening than for subsequent cycles (due to nut embedment and tensioner alignment improvement between cycles). This means that if the assembly procedure does not include 2 to 3 nut tightening and pressure cycles at each pass, then the achieved final bolt load will be significantly less accurate. The overall results for NLLF testing (Fig. 6) show a comparison between various tests results and theoretical predictions using Eq. 1. It can be seen that in most cases, the agreement is very good. However, in the case of the 1.5 inch diameter bolt, the NLLF appears to be under-predicted by the theory. However, based on other additional test results, this was thought to be primarily due to unaccounted for additional flexibility in the test set-up, rather than a problem with the theory. The other interesting point to take from this graph is that the NLLF increases significantly as the bolt length to diameter ratio reduces below a value of around 5. In addition, the greater slope in the small L/d region of the graph (i.e.: short bolt region) indicates that it will be much more difficult to accurately predict the actual NLLF for a short bolt as the value is becoming significantly more variable. Practically, if these results are combined and we look at natural variance and the variance associated with both bolt stress and bolt length, it means that it will not generally be possible to predict the NLLF with a high level of accuracy and that the ±1% quoted for tensioner accuracy is probably best case. In particular, hydraulic tensioning will be very inaccurate for low stress levels and relatively short bolts. However, that said, so will hydraulic torque, so therefore this is something that the joint designer ought to be aware of, so that designs do not have short bolts or excessive bolt area. CONFIRMING THE THEORY Once the nut rigidity was established, which varied between bolt sizes tested but was found to be acceptably approximated by a value of 3.5E-7 mm/n, the next step was to demonstrate that the method could be applied to accurately predict both the NLLF and BLLF on actual standard piping joints. For these tests, several different size and classes of joints were assembled using hydraulic tensioners. The majority of the joints were raised-faced with spiral wound gaskets, although for comparison purposes, a RTJ joint was also assembled. In addition, different bolt materials were used during the assembly such that ability to tighten to values approaching or possible effects of bolt yield during tightening could also be assessed in practical terms. The tests consisted of measuring bolt load by both a load cell mounted on one bolt and, in addition, by either manual/ automated elongation measurement of the bolt elongation (for B8M bolts) or ultrasonic measurement of bolt elongation (for B7 bolts) for the remaining bolts. An example of the test setup is shown in Fig. 7 and Fig. 8. The joint was assembled using 5% tensioner coverage, with the bolt load measured at each stage of the tightening process. The primary goal of the testing was to demonstrate the accuracy of the proposed elastic interaction method in predicting the NLLF and BLLF, therefore the tests were generally stopped after only two passes (since those passes have the highest levels of interaction occurring) and no attempt was made to continue until an accurate load was achieved. However, in some cases additional passes were made, in order to demonstrate that the method was valid for more than two passes. Comparison of the theory and a selection of the test results for both NLLF and BLLF are shown for various joint/bolt combinations in Fig. 9 to Fig. 21. These figures present the average residual bolt stress after each tightening pass for both the first half of the bolts (A bolts) and the second half (B bolts). was performed to the bolt stress level indicated by the ratio listed in the figure title. The NLLF factor is confirmed by the reduction in bolt stress from that applied by the tensioner to the final residual bolt stress and the BLLF is confirmed by the final bolt stress for the A bolts after the tightening is complete. It can be seen that in general the accuracy of the theory presented in this paper was very good. There was some deviation between theory and test for the B8M trials, but this was thought to be primarily due to the difficulties in measuring the residual bolt load for the bolts once above. This can be confirmed 3 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

by the fact that several of the test measurements resulted in apparent stress levels above that applied by the tensioner (i.e.: a NLLF in excess of 1), even once adjusted for yield elongation. Since increasing load on removal of the tensioner is not possible, this simply indicates the difficulty in measuring bolt load. One important aspect to note from these trials is that the use of a load cell to monitor the NLLF and BLLF was not possible for most cases studied. This is due to the additional bolt length required for the load cell, which changes the resulting NLLF and BLLF sufficiently that the results from the load cells, while indicative of the general magnitude of the load change were not sufficiently accurate to predict the actual level of NLLF or BLLF. PRACTICAL APPLICATION There are several practical applications of the NLLF and BLLF theory that are extremely useful in the field. The first of these is for assembly of the bulk of the standard piping joints with B7 bolts and spiral wound gaskets. In these cases, it is typically possible to achieve accurate assembly using just two passes. In order to determine the amount of over-tension required for each pass, the values of R1 and R2 must first be determined (Table 1 and Table 2). Once these values are known, then the overpressure for and can then be determined (Table 3 and Table 4). Therefore, these last two tables can be used directly in the field to determine the required hydraulic pressure for both and tensioning assembly of standard flanges. The variation in the values is worth noting (particularly by comparison to current guidance which tends to specify around a 25% increase between to pressure). The second practical application of the NLLF and BLLF theory is in determining the required number of passes or the optimal hydraulic pressures for more difficult joints to assemble, such as RTJ joints. The RTJ joint is significantly more difficult to assemble using 5% or less hydraulic tensioner coverage, since there is a lot of flange movement related to gasket compression. This means that the R2 factor is significantly higher for RTJ joints, making it difficult to achieve an accurate assembly with only two passes. This can best be illustrated by the test example of assembling a DN 2 (NPS 8), class 9 joint to a target bolt stress of 265 MPa. The assembly results for a standard assembly using approximately 25% over-load for compared to is shown in Fig. 2. The NLLF and BLLF theory is also plotted on the graph. The test involved a total of 5 assembly passes on the joint and, while there is definite improvement in average bolt load with each pass, it can be seen that even after 5 passes the average bolt load is still only at 84% of the target and the spread between the bolts is 21% of the target (i.e.: still unacceptable). If the pass over-load is adjusted in line with the theory, then it can be seen that after only 3 passes the target bolt load is achieved accurately (Fig. 21). Of course, this requires a significantly higher bolt stress, but since most joints using RTJ gaskets do not require a high target bolt stress then in a lot of cases this will be an acceptable practice. Therefore, by using this approach, it will be possible to both reduce the number of passes required and achieve a more accurate final bolt stress. The third practical application is in determining the number of passes (and associated hydraulic pressure) for an application that has a maximum limit for the bolt stress that is close to the target bolt stress. An example of this case might be on a joint with B8M class 2 bolts, where it was considered desirable to limit the bolt stress during to below. For example, we will look at the case used in Noble [1], which was a NPS 1 (DN 25), class 9 joint with a target assembly bolt stress of 3 MPa. The bolts for that joint size are 1-3/8, so therefore the is 345 MPa. Therefore, if limiting the bolt stress to, the maximum allowable over-load on would be 345/3 = 1.15. Since this is only just more than the over-load required to counteract the NLLF, it will need to be applied in full at each pass. In that case, the resulting bolt stress after each pass is: Residual Stress Bolts Residual Stress Bolts After and After Pass C After Pass D 168 MPa 36 MPa 258 MPa 36 MPa 252 MPa 36 MPa It can be seen that the progress towards 3 MPa bolt stress becomes very slow, however even after only 4 passes (two tightening actions per bolt), the average bolt load is within 1% of the desired target bolt stress (282MPa = 94%) and the spread between the bolt load is relatively small (i.e.: within the stated nominal accuracy of the method). If the overload is allowed to be slightly higher, for example 11% of, then convergence is improved greatly. In that case, if is performed at the maximum permissible 1.26, at 1.25 and Pass C at 1.14 then after the three passes the average bolt load will be 98% of target (294 MPa) and the variation will only be 3% (in theory). Therefore, a relatively small increase in maximum permissible bolt stress, or relatively small decrease in target bolt load will allow the accurate use of tensioners for this application. TIGHTENING TO NEAR YIELD From both the theory and the test results, where B8M bolts were assembled to levels greatly exceeding and even exceeding the material certificate yield value, it is evident that it is certainly possible to assemble to values near. In some cases, it may be necessary to exceed in order to establish the target bolt stress in relatively few passes, 4 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

but there is nothing preventing obtaining the desired bolt stress level. In fact, there is really no impact on the tensioning procedure, since all plasticity occurs during the initial bolt stress applied by the tensioner and so the subsequent bolt behavior once the tensioner is released is purely elastic. Of course there are limits as to the bolt stress that ought to be applied during tensioning. There are two primary reasons for those limits; not wanting to change the bolt properties and secondly, not wanting to assemble to a level where component failure may occur. The first requirement (to not change the bolt hardness by strain-hardening) is only applicable in cases where it was necessary to prevent stress corrosion cracking (SCC). This would, therefore, commonly only be a consideration for A193-B7M and A193-B8.B8M class 1 bolts. It may also be a consideration for some higher alloy bolts in SCC service. It is not a consideration for A193-B8M class 2 bolting, for example, since that material is not used in SCC environments. For bolts in SCC service, it is perceivable that additional strain applied during assembly may bring the material hardness above the NACE MR-175 / MR-13 limits for the material. In that case, the bolts may be subject to stress corrosion cracking during operation which may result in catastrophic failure of the bolts, which is certainly not desirable. However, two things must be kept in mind with regards to this. The first is that the actual yield of the bolt will be likely to significantly exceed the value (by 2% or more). Secondly, although minor yielding occurs below the.2% yield value (as discussed in Noble [1]), the level of strain hardening associated with that yield is extremely small. In addition, it only occurs once during the assembly and does not increase progressively with subsequent assembly cycles. It is therefore not a case of the hardness increasing incrementally once or even yield is passed. In fact, the strain hardening accumulation and associated increase in hardness of the material is rather slow. The case in point of this is the B8M class 1 bolts that were assembled to around two times the value during the testing detailed in this paper. Prior to assembly it can be seen that the actual yield was almost double, with a compliant hardness (so these bolts would not have seen any hardening at tensioner load levels). After testing, where actual yield was exceeded, three of these bolts were then subsequently cross-sectioned and micro-hardness values taken at the thread root region. In two out of the three cases, the average hardness actually reduced in the strained region of the bolt compared to the hardness in a region without plastic strain. In addition, the increase in the hardness of the third bolt was small (within the accuracy/repeatability of the measurements). This means that, while caution is required with tensioning B7M or B8M class 1 bolts, it is not necessary to be overly conservative. The risk associated with the potential for increasing the bolt hardness must be weighed against the risk of joint leakage associated with low bolt loads (particularly with B8M class 1 bolts). The second consideration, to avoid joint component damage, is primarily only really a concern with respect to the bolts. With 5% tensioner coverage, the over-load will not over-stress the flanges or gasket, since only half the bolts are being loaded to that level. In addition, as the pressure is applied, the bolt load is reduced, so therefore the actual load seen by the flanges and gasket will only be slightly more than the target bolt stress and certainly within the nominal tolerances of the flange or gasket strength limits. So, we are only concerned with bolt failure or nut stripping. In that case, particularly in non-scc services were some minor strain is tolerable, we are then only concerned with the potential for mechanical failure. This can be best examined by looking at the ratio of the and Specified Minimum Tensile Strength (SMTS) of several common bolt materials, as below: Material (diameter) SMTS Ratio (MPa / ksi) (MPa / ksi) A193-B7 (< 2.5 ) 72 / 15 86 / 125.84 A193-B7M (< 4 ) 55 / 8 69 / 1.8 A193-B16 (< 2.5 ) 72 / 15 86 / 125.84 A193-B8M cl. 1 (all dia.) 25 / 3 515 / 75.4 A193-B8M cl.2 (1.5 ) 345 / 5 62 / 9.56 A193-B8M cl.2 (1.25 ) 345 / 65 655 / 95.68 A453-Gr66 (all dia.) 585 / 85 895 / 13.65 The first thing to note from this comparison is that the is around 8% of SMTS for the B7 and B16 bolts. This indicates that if the bolt stress approaches during Pass A tensioning, that there is less than a 17% buffer on ultimate failure for B7 bolts. Given the inaccuracies with applying tensioner pressure, this would probably be considered insufficient buffer against failure for most day-to-day applications. Therefore, for B7 and B16 bolts, it is advisable to limit the maximum bolt stress to below and, potentially even lower than that in order to create a greater buffer against bolt failure during tensioning. Conversely, the 1-1/2 inch, B8M cl.2 bolts have a that is only around 55% of their SMTS. Therefore, there would still be a significant buffer against bolt failure even if the bolt stress levels exceeded by up to 2%. However, for other sizes of B8M cl.2 bolts, the ratio of to SMTS is lower and, in addition, the ultimate nut strength for those bolts is only required to correspond to a bolt stress of 55MPa (8 ksi). Therefore, tightening to 12% of for a 1.25 inch B8M cl.2 bolt would be at the minimum ultimate strength of the nuts and therefore nut failure may be possible. Therefore, for all sizes of B8M cl.2 bolts it is necessary to limit the Pass A maximum bolt stress to below a nominal value of, for example, 45 MPa (65 ksi), which offers a 2% buffer against minimum specified nut failure strength. 5 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

CONCLUSIONS The analysis and test results presented in this paper allow the following conclusions to be made: The accuracy of tensioners at low bolt stress levels and for short bolts (L/d less than 5) will be greatly reduced. This should be considered during tensioner hydraulic pressure selection. However, ideally bolts with L/d less than 5 would be eliminated at the design stage. The L/d ration for standard B16.5 and B16.47A flanges is shown in Fig. 22, for information. It can be seen that it appears that many of the joints have L/d ratios below 5. However, if only the joints that are likely to be tensioned are examined (Fig. 23), then it can be seen that the lowest values are only just below 5. However, based on the test results, this implies that at least a percentage of standard flanges should ideally have the NLLF further increased to account for the effect of low L/d ratio. Hydraulic Tensioner procedures should include 3 pressure cycles during each pass (prior to removal of the tensioners) such that the nut is properly embedded and the NLLF minimized. Failure to cycle the tensioner will result in loss of accuracy. Load cells should not be used to measure NLLF or BLLF and elongation methods should consider residual bolt elongation (bolt yield). Both methods can, therefore, be inaccurate when determining residual bolt load under certain circumstances. When tightening standard flanges with spiral wound gaskets, the pressure multiplier values listed in Table 3 and Table 4 may be used to directly establish the required hydraulic pressure. For RTJ joints and other joints with gaskets that exhibit significant movement during assembly, the target pressures must be adjusted significantly from those used for spiral wound gaskets. In many cases a multi-pass procedure will be required. The pressure multiplier can be accurately determined using the same theory as that outlined in this paper. The maximum bolt stress during should be checked against bolt. If the value exceeds 85% for bolts other than B8/B8M, then a three Pass (or more) procedure must be utilized with the pressure set below 85% of. For B8/B8M bolting, the bolt stress during should not exceed the lesser of [a multiplier on (between 1. and 1.2), and 45 MPa (65 ksi)]. Once again, if the bolt stress exceeds those values then the pressure would be adjusted down and a multi-pass procedure employed to achieve the desired target bolt stress. Alternatively, 1% tensioning should be considered, if feasible. In some cases, it may not be possible to accurately obtain the desired bolt stress using tensioner assembly (for example, for spiral wound gaskets, when the maximum permissible bolt stress multiplied by the R1 factor listed in Table 1is less than the target bolt stress). Even with the additional accuracy available using these methods, it is still advisable to perform checkpasses until no further nut turn is obtained at the final pass pressure. This allows for the possibility of variance in joint component behavior and possible problems during assembly of the joints. It has been shown that the current practice of using only a pressure and then a pressure multiple times (until the nuts do not turn) is not a very effective way of reaching the desired target bolt stress. The overall gain in bolt stress for the third and subsequent passes is very small. A much simpler approach is to utilize a continually varied pressure for each pass, with the pressures selected using the theory outlined in this paper. Using the example RTJ assembly shown in this paper, it was demonstrated that it can reduce the number of passes required from in excess of 7 to only 3. An alternative would be to use 1% tensioning (which eliminates the BLLF interaction). It should be noted that many of the concepts outlined in this paper are also applicable to torque assembly as well. By applying the same methods, an estimate of the required number of passes may be obtained and better idea of the residual accuracy of the method is also possible. In some cases, it would also be possible to increase the intermediate pass torque values in order to greatly reduce the number of passes required (and at the same time improve accuracy). It is also worth noting that while torque assembly may appear superior for assembling B8M bolts (since no overload is nominally required), that is not actually the case, since due to the torsion stress that is generated by the method, the actual yield state of the bolt during application of torque is likely to be similar to that generated during tensioning (in terms of generating strain hardening or initiating bolt yield). REFERENCES [1] EN1591-1, Flanges and their joints. Design rules for gasketed circular flange connections, EN Standard [2] Tentec method for calculating load loss factor, located at: http://www.tentec.net/tips.htm [3] SPX method for calculating load loss factor, located at: http://www.spx.com/en/bolting-systems/resources/bolt-stressgraphs/ [4] Noble, R., 213, Working Bolts Near to Yield Theory, Experience and Recommended Practice, ASME PVP, Paris, France, PVP213-97956 [5] Brown, W., 1993, Design and Behaviour of Bolted Joints 3rd International Conference on Fluid Sealing, CETIM, Nantes, France, pp. 111-121 6 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

Table 1 R1 value for Standard Flanges, SPWD Gasket Tensioner-to-Nut Load Loss Ratio (R 1) NPS 15 3 6 9 15 25 ASME B16.5 Flanges 3.92.93.92.91.87.85 4.95.93.91.89.85.82 5.93.93.9.87.83.79 6.92.94.91.89.86.76 8.92.93.9.88.83.78 1.93.92.9.89.8.74 12.93.92.91.9.81 14.93.94.91.88.79 16.94.93.9.86.78 18.91.92.87.84.75 2.92.93.89.83.74 24.92.92.87.79.71 ASME B16.47 Series A Flanges 26.97.96.92.79 28.98.96.91.79 3.98.95.89.78 32.98.94.88.77 34.98.94.86.75 36.97.94.85.76 38.98.91.83.71 4.98.91.82.71 42.97.9.8.71 44.98.87.8.7 46.98.85.8.68 48.98.89.79.68 Table 2 R2 value for Standard Flanges, SPWD Gasket Bolt-to-Bolt Load Loss Ratio (R 2) NPS 15 3 6 9 15 25 ASME B16.5 Flanges 3.22.41.28.28.27.16 4.46.37.28.36.26.2 5.46.31.31.35.24.19 6.39.42.33.3.21.17 8.34.43.33.37.23.17 1.51.42.4.4.23.17 12.47.48.43.41.26 14.54.56.48.39.29 16.6.56.49.37.31 18.48.53.41.38.29 2.54.54.47.35.3 24.57.59.47.37.33 ASME B16.47 Series A Flanges 26.81.78.62.44 28.84.77.63.49 3.82.75.55.45 32.85.75.58.45 34.86.73.53.48 36.83.76.56.49 38.85.45.34.34 4.85.47.31.33 42.82.44.32.31 44.84.34.31.32 46.82.33.3.32 48.83.47.33.29 Table 3 Pressure Multiplier, SPWD Gasket Pressure Multiplier NPS 15 3 6 9 15 25 ASME B16.5 Flanges 3 1.33 1.51 1.39 1.41 1.45 1.37 4 1.53 1.47 1.41 1.53 1.48 1.46 5 1.56 1.41 1.45 1.55 1.49 1.5 6 1.5 1.5 1.46 1.46 1.41 1.54 8 1.45 1.53 1.49 1.57 1.49 1.5 1 1.62 1.54 1.55 1.57 1.53 1.58 12 1.59 1.61 1.57 1.56 1.55 14 1.65 1.67 1.63 1.57 1.63 16 1.7 1.69 1.66 1.58 1.68 18 1.64 1.66 1.62 1.63 1.72 2 1.67 1.66 1.65 1.64 1.75 24 1.71 1.73 1.68 1.72 1.75 ASME B16.47 Series A Flanges 26 1.86 1.86 1.77 1.82 28 1.87 1.85 1.79 1.89 3 1.86 1.85 1.74 1.87 32 1.9 1.86 1.8 1.89 34 1.9 1.84 1.77 1.97 36 1.88 1.87 1.83 1.96 38 1.89 1.59 1.62 1.9 4 1.89 1.62 1.6 1.88 42 1.87 1.59 1.65 1.85 44 1.88 1.54 1.63 1.9 46 1.87 1.56 1.62 1.93 48 1.87 1.65 1.68 1.91 Table 4 Pressure Multiplier, SPWD Gasket Pressure Multiplier NPS 15 3 6 9 15 25 ASME B16.5 Flanges 3 1.9 1.7 1.9 1.1 1.15 1.18 4 1.5 1.7 1.1 1.13 1.17 1.22 5 1.7 1.8 1.11 1.15 1.21 1.26 6 1.8 1.6 1.1 1.12 1.17 1.31 8 1.9 1.7 1.11 1.14 1.21 1.27 1 1.7 1.8 1.11 1.13 1.24 1.35 12 1.8 1.9 1.1 1.11 1.24 14 1.8 1.7 1.1 1.13 1.26 16 1.6 1.8 1.11 1.16 1.29 18 1.1 1.8 1.15 1.18 1.33 2 1.8 1.8 1.12 1.21 1.36 24 1.9 1.9 1.15 1.26 1.45 ASME B16.47 Series A Flanges 26 1.3 1.5 1.9 1.26 28 1.2 1.5 1.1 1.27 3 1.2 1.6 1.12 1.29 32 1.3 1.6 1.14 1.31 34 1.2 1.6 1.16 1.33 36 1.3 1.6 1.17 1.31 38 1.2 1.9 1.21 1.41 4 1.2 1.1 1.21 1.41 42 1.3 1.11 1.25 1.42 44 1.2 1.15 1.25 1.44 46 1.3 1.17 1.25 1.46 48 1.2 1.12 1.27 1.48 7 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

Remaining Load Factor (Applied/Residual).9.8.7.6.5.4.3.2.1. q N q f q g Figure 1 Spring in Series Model of the Joint Bolt tensioned at opposite end to load cell. Nut tightened against Spacer length varied in order to change L/D ratio Figure 2 Nut Rigidity Test Setup Yield from Test Cert. 1 2 3 4 5 Applied Bolt Stress (MPa) Cycle#1 Cycle#2 Cycle#3 Figure 3 Example #1 NLLF vs. Bolt Load, 1-1/8 B8M cl.1 q b L b Load cell connected to a datalogger L/d = 5.6 Remaining Load Factor (Applied/Residual).9.8.7.6.5.4.3.2.1. Yield from Test Cert. 1 2 3 4 5 Applied Bolt Stress (MPa) Cycle#1 Cycle#2 Cycle#3 Cycle#4 Figure 4 Example #2 NLLF vs. Bolt Load, 1-1/8 B8M cl.1 Remaining Load Factor (Applied/Residual Load Loss Factor (Applied/Residual).9.8.7.6.5.4.3.2.1. Yield from Test Cert. 1 2 3 4 5 6 7 Applied Bolt Stress (MPa) Cycle#1 Cycle#2 Cycle#3 Cycle#4 Figure 5 Example NLLF vs. Bolt Load, 1-1/8 B8M cl.2 1.9.8.7.6.5.4.3.2.1 L/d = 5.6 L/d = 5.6 Test: 1.125" B7 Test: 1.125" B8Mcl.2 Test: 1.125" B8Mcl.1 Test: 1.125" B8Mcl.1 Test: 1.5" B7 2" B8Mcl.2 B7 1.125" Theory B7 1.5" Theory B7 2" Theory B7 1.625" Theory 5 1 15 2 25 Length/Diameter Ratio Figure 6 Comparison NLLF Theory vs. Test Results 8 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

Automated Elongation Measuremen Load Cell 6 5 4 3 2 1 (Sy actual = 62 MPa) Figure 1 DN4 (NPS16), cl.6, SPWD, B8M2, 1.6 Figure 7 DN 4 (NPS 16), cl.6 test setup 25 2 15 1 5 Figure 11 DN35 (NPS14), cl.15, SPWD, B7,.44 Figure 8 DN 2 (NPS8), cl.9 & DN 35 (NPS14), cl.15 4 35 3 25 2 15 1 5 (Sy actual = 62 MPa) Figure 9 DN4 (NPS16), cl.6, SPWD, B8M2, 1.2 35 3 25 2 15 1 5 Figure 12 DN35 (NPS14), cl.15, SPWD, B7,.63 9 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

3 Sy actual = 375 MPa 25 2 15 1 5 Figure 13 DN3 (NPS12), cl.3, SPWD, B8M1, 1.5 7 6 5 4 3 2 1 Sy actual = 62 MPa Figure 16 DN3 (NPS12), cl.3, SPWD, B8M2, 1.2 5 45 4 35 3 25 2 15 1 5 Sy actual = 375 MPa Figure 14 DN3 (NPS12), cl.3, SPWD, B8M1, 2.1 5 45 4 35 3 25 2 15 1 5 (Sy actual = 62 MPa) Figure 17 DN2 (NPS8), cl.9, SPWD, B8M2, 1.2 4 35 3 25 2 15 1 5 5 45 4 35 3 25 2 15 1 5 (Sy actual = 414 MPa) Sy actual = 414 MPa Figure 18 DN2 (NPS8), cl.9, SPWD, B8M2, 1.55 Figure 15 DN3 (NPS12), cl.3, SPWD, B8M2, 1. 1 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms

45 8 4 7 Bolt Stress (MPa) 35 3 25 2 15 1 5 Bolt Length to Diameter Ratio 6 5 4 3 2 1 cl.15 cl.3 cl.6 cl.9 cl.15 cl.25 4 8 12 16 2 24 28 32 36 4 44 48 Figure 19 DN2 (NPS8), cl.9, SPWD, B7,.66 Flange NPS Bolt Stress (MPa) Bolt Stress (MPa) 35 3 25 2 15 1 5 Pass C Pass D Pass E Figure 2 DN2 (NPS8), cl.9, RTJ, B7,.5 5 45 4 35 3 25 2 15 1 5 Pass C Figure 22 B16.5/B16.47A Bolt Length to Diameter (L/d) Ratios, All Joints NPS.5 to NPS 48 Bolt Length to Diameter Ratio 8 7 6 5 4 3 2 1 4 8 12 16 2 24 28 32 36 4 44 48 Flange NPS cl.15 cl.3 cl.6 cl.9 cl.15 cl.25 Figure 23 B16.5/B16.47A Bolt Length to Diameter (L/d) Ratios, Joints Commonly Tensioned Only Figure 21 DN2 (NPS8), cl.9, RTJ, B7,.7 11 Copyright 214 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/214 Terms of Use: http://asme.org/terms