Presenting facts – How exactly should equipment flanges be entered into calculations?

Thread Content

Calculation of equipment flanges: a well-known yet still controversial issue. As is well known, our **standard specifies the use of standard equipment flanges, and it is clearly stipulated in the standard that the use of such standard flanges exempts one from the calculations required by GB150. ……As a result, for a long time, standard equipment flanges did not require any calculation, and years of practice have shown that using standard flanges directly poses no problems. Although practice is the criterion for testing truth, I don’t know since when standard equipment flanges have had to be calculated using SW6. I guess one reason might be that this SW6 software comes with a module for flange calculations; all that’s needed is to enter a few numbers to carry out the calculation. Without such software and if one had to do the calculations by hand, I think most people wouldn’t bother with it and would simply choose standard flanges to meet the requirements of the standards. In case there are any issues, responsibility can be traced back to the standards, and the person in charge of design review can’t be held accountable, thanks to the support of these standards. Well, since we have to use SW6 for the calculations now, then let’s go ahead and do them. The prerequisite for carrying out these calculations is that we first need to understand the theoretical foundations and calculation models related to flanges. But in reality, SW6 provides clear instructions regarding the input of each parameter; I just need to enter a few values and the calculation results are obtained right away – there’s no need to understand any theories or calculation models. Well, the onlookers think it’s no big deal to just watch what’s happening; so they proceed by entering the values according to the parameters in the software. And that’s how they enter a series of numbers into the software based on the dimensions specified in the standards. Here’s an example: for a standard flange of DN900PN=1.6Mpa, a 3mm corrosion allowance is taken into account. The data is entered quickly; a 3mm corrosion allowance is also applied to the inner diameter of the flange as well as to the thicknesses of the larger and smaller ends of the conical neck. After clicking ‘Calculate’, the results show that all three stress checks pass, and the stiffness requirements are met. Even after taking into account the 3mm corrosion allowance, the calculations yield satisfactory results. We used a standard flange, and we also conducted the calculations using SW6 – all results were acceptable. Perfect, no problems at all! ……Thus, for a period of time, a method for calculating standard equipment flanges was used, based on the parameters mentioned above. Years of practice have shown that using standard flanges along with the calculations outlined above is effective, and further years of experience have confirmed that there are no issues with standard equipment flanges. Although practice is the criterion for testing truth, at some point a dispute arose regarding the calculation of flanges for standard equipment. It is easy to see that in the aforementioned calculations, the effective thickness g0 of the smaller end of the neck is determined by subtracting 3 mm for corrosion allowance from the actual size of the standard equipment flange (that is, 16–3=13). This is where the dispute lies. One view holds that the thickness of the smaller end of the conical neck is calculated incorrectly; it should not be based on the actual size, but rather on the thickness of the connected cylinder minus the corrosion allowance (that is, 10–3=7). Using the standard thickness values is unsafe. Well, then let’s calculate it based on the cylinder thickness minus the corrosion allowance. Enter 7 mm as the thickness of the smaller end of the conical neck to carry out the calculation; clicking ‘Calculate’ gives us the following result: The bright red font clearly indicates that the verification has failed. If we compare the values, the difference is quite significant – especially since the stiffness index J=1.60, which is 60% higher than the acceptable value. That’s a very large difference, and it really confuses me as a junior designer. This is absolutely outrageous; what should I do? Let’s first take a look at the comparison of results: the axial stress σH increased from 177.66 Mpa to 213.09 Mpa. It was quite close – the axial stress barely passed the verification criteria ； The tangential stress σT and radial stress σR also change significantly, but these two stresses are relatively small; there are basically no cases where the requirements are not met, so they can be ignored for now. So what to do if the stiffness does not meet the specifications? All that can be done is to adjust a few input parameters. The thickness of the smaller end of the neck cannot be changed; either the thickness of the larger end of the neck must be adjusted, or the height of the neck, or the thickness of the flange. But I don’t know which parameter should be adjusted to meet the requirements, so I have to try adjusting each parameter one by one. Surprisingly, the software suggested that increasing the flange thickness to 71 mm would suffice to meet the stiffness requirements. Well, then why go through the hassle of adjusting each parameter individually? I can just increase it by 1 mm to get an even number of 72 mm – this will still satisfy the calculation requirements. By changing just one parameter, it’s possible to indicate this change in the drawings. If too many parameters are adjusted, it becomes difficult to determine the optimal structure, and it’s possible to end up with a non-standard flange, which only creates more trouble for oneself. Well, if adjustment isn’t possible, then I’ll just increase the thickness directly; as long as the calculations work out fine. But this increase raises the thickness from 52 mm, the standard value for flanges, to 72 mm – an increase of 20 mm in total. At that moment, I was secretly relieved that it wasn’t at the manufacturing facility; if it had been there, I definitely wouldn’t have kept my job. ……Thus, for a while there was a practice of using the thickness of the docking cylinder as the standard for calculating the thickness of the small end of the cone neck in equipment flanges, which resulted in these standard flanges being altered significantly – their thicknesses were increased repeatedly or they became non-standard flanges. But why wasn’t there any problem when using standard flanges without such calculations, while now all calculations fail? It will likely remain a question in many people’s minds for a long time. For many years, adhering to the principle of safety first, we have used the method of entering data based on the thickness of the smaller end of the cylinder, and we have never had the slightest doubt about this so-called conservative approach. The reason we are discussing this topic today is because it has been brought up again and again, with various opinions expressed; after all, this has always been how we have done things in the past, and no problems have arisen as a result ； For example, this is the rule within our organization – rules are rules, and it’s right to abide by them; there’s no need to worry about it. Well, it seems like everything makes sense, but... I’m confused again. As a result, a perfectly normal standard flange gets transformed into something unrecognizable, leaving the designers at a loss. God, please save me! Just ignore it; follow the review suggestions. Do whatever the reviewers ask. Yeah, that’s right; it’s correct. After all, it’s the three of us who are in charge of reviewing it, and if we all think it’s fine, then it’s okay. Until one day, I came across the requirements of a company regarding the calculation of flanges for standard equipment. After comparing those requirements with my own, I realized that their demands were quite reasonable, while the results obtained using my own calculation method turned out to be extremely conservative. The main reason for this was that the calculation model, which used the thickness of the cylinder body as the thickness of the narrow end of the cone, differed greatly from the actual flange geometry, resulting in highly conservative values (especially in terms of stiffness, where the differences were significant). It’s fine to be conservative, but when the stiffness is already sufficient, wasting material to increase it any further can’t simply be described as conservatism – it’s actually incorrect. Using such incorrect calculations to guide design only leads to further mistakes. So today, no theory – just facts, simple and clear for easy understanding. Let’s use the previous example for comparison: Fact 1: The calculation model that uses the thickness of the connecting cylinder as the thickness of the smaller end differs significantly from the actual structure. (a) The actual standard flange model; (b) The flange model with the red-filled portion removed. If both of these structures are entered into the software using the thickness of the connected cylinder as the thickness of the conical neck’s smaller end, then their inputs in the software will be identical, and the calculation results will also be the same. However, the actual structures differ greatly. It is clear that structure (b) lacks the red-filled portion present in structure (a), and it’s not just a small area that’s missing – it’s an entire ring of structure. With so much of the load-bearing structure missing, the axial force increases, and the stiffness naturally decreases significantly. It is easy to see at this point that if a calculation model is used in which the thickness of the cylinder body is taken as the thickness of the smaller end, it is actually the model from structure (b) with the red-filled portion removed; such calculations yield overly conservative results. Fact 2: Is it really unsafe to enter the thickness of the smaller end of the conical neck based on its actual structure for calculations? (c) Enter the model using the thickness of the smaller end of the cone of the standard flange. If SW6 is calculated using the actual thickness of 16–3=13 mm for that smaller end, then the calculation model will be consistent with the actual structure. However, the axial stress calculated in this way corresponds to the axial stress in the blue-colored section shown in the diagram, that is, the axial stress based on a thickness of 13 mm, rather than the axial stress at the section with a thickness of 10–3=7 mm at the end of the cylinder. This is why many people consider the calculations to be unsafe, as the axial stress at the end of the cylinder is much higher than that at the smaller end of the flange’s cone; axial stress is primarily caused by edge bending moments, and it is inversely proportional to the square of the wall thickness. Indeed, there is no doubt about this: the axial stress at the cylinder end must be high. However, exactly how much higher it is is a matter of debate. We certainly took this issue into consideration when formulating our standards. In ASME standards, flanges do not have any straight edges, whereas our standards require such straight edges, and there are clear specifications regarding the minimum thickness of the butted cylinders. When these minimum thickness requirements are not met, it is possible to increase the height of the flange, thereby increasing the length of the straight edge. The purpose of this straight edge is to take into account the pattern of stress reduction at the edges. As is well known, the axial stress in calculations is the edge stress generated by edge moments, and this edge stress has two notable characteristics: locality and attenuation. The rate of attenuation of edge stress is very fast, with the stress decreasing to a minimal level far from the edges. Therefore, the question arises as to by how much the edge stress can be reduced by the length of this straight edge section; this is indeed something we cannot determine. Moreover, the rate of attenuation of bending stress is slower compared to that of film stress. Well, if we consider (d*σn)^0.5 = (900*16)^0.5 = 120 mm, then indeed the length of the straight edge section of the standard flange is not sufficient to ensure a sufficient reduction in axial bending stress; it’s therefore impossible to determine how much axial bending stress remains once reaching the cylinder’s cross-section. If, in addition, the axial membrane stress generated by internal pressure is taken into account, could this lead to high stress levels? Here, a method of judgment that is not entirely precise but reasonable is recommended: in this case, change the thickness at the minor end to the thickness of the input cylinder and carry out further calculations (i.e., using model (b)). The axial stress calculated in this way will be much higher and more conservative, but in most cases the checks can still be passed. As shown in the calculations above, 213.09 Mpa is still less than 253.80 Mpa; thus, the stiffness meets the requirements based on the actual structure, and the axial stress at the cylinder ends will also satisfy the criteria. Therefore, this standard equipment flange is very safe, and there is no need for unnecessary adjustments such as increasing the flange thickness. Of course, wealthy companies don’t need to consider it. Fact 3: An intermediate calculation method that is also conservative, but to a much lesser extent than the method in which the thickness of the cylinder body is used as the thickness of the smaller end. In model (d), the slanted lines of the standard flange cone section are extended until they intersect the cylinder body. Of course, if the axial stress of the standard flange, calculated using model (b), does not meet the required standards and there is a lack of numerical data for evaluation, or if some people still feel uncertain due to conservatism, then the intermediate calculation model shown in figure (d) can be used. By comparing the three calculation models (a), (b), and (d), it is clear that the calculation model (d) has less red-filled area compared to the actual model (a), while it has more blue-filled area compared to model (b). When using the parameters of this model for calculations, the thickness of the smaller end remains 7 mm; only the height of the cone neck needs to be changed from 35 mm to 56 mm. The calculation results are shown in the figure below. Although these results still do not meet the requirements, it is evident that J = 1.17 > 1.0, meaning the stiffness value has increased significantly. The axial stress, calculated using the thickness of the cylinder body as the thickness of the cone neck’s smaller end, is σH = 180.85 MPa, which is very close to the value of 177.66 MPa obtained from the actual flange model (a), and it is much lower than the value of 213.09 MPa from model (b). At this point, it would be sufficient to adjust the flange thickness to 60 mm, which is 11 mm less than the 71 mm required by model (b). Although the stiffness check in this example still fails, it is possible that other flanges might meet the requirements in this case, in which case no further adjustments to any parameters would be necessary. It is also easy to see that, in fact, the calculation model (d) can be used as a method to verify whether the axial stress of the standard flange is within acceptable limits, and it is more convincing than the calculation model (c): if the axial stress calculated using (d) is within acceptable limits, then it can be assumed that the axial stress of the actual standard flange (a) is also within acceptable limits, and its stiffness can also be confirmed to be satisfactory based on the calculations using the actual model. In summary, with the three calculation models, (a) + (c) + (d) = 99.9% compliance, it can be determined that the flange of the standard equipment is fully compliant, and no unnecessary modifications are required. 1. (a) The calculation model is sufficient to ensure that the stiffness of the flanges on standard equipment meets the required standards; (c) + (d) ensure that the three stress values are within acceptable limits, i.e., that the strength requirements are met. The reason it cannot be 100% certain is that the edge moments in the aforementioned three models are not exactly the same, and therefore the calculation models based on (c) + (d) cannot cover model (a) 100% completely, which is why a margin of 0.1 is left. 2. The calculation method (b), which uses the cylinder thickness as the input for the thickness of the small end of the cone, is too conservative and represents the least desirable approach. As can be seen from the calculations, the main reason for failure in these calculations is insufficient stiffness; since the stiffness index J is inversely proportional to the square of the thickness of the small end, a decrease in thickness leads to excessive deviations in the calculated value of J. Yet the actual stiffness of the flange is fully satisfactory. If this calculation model is used to increase the flange thickness or adjust other parameters in order to meet the stiffness requirements, and if erroneous calculation results are then used to guide the design, wouldn’t that be adding error on top of error? 3. This article only provides an example of a flange for standard equipment to offer some ideas and methods; specific situations should be analyzed on a case-by-case basis. 4. Facts speak louder than words; no theoretical explanation is needed. Just a few sets of comparisons are sufficient to prove the facts, and I believe my friends will be able to see them clearly and plainly. Furthermore, this article represents only personal views; due to limited knowledge, any inaccuracies or incomplete considerations are welcome to be pointed out and corrected. Welcome to search for and follow the WeChat official account \"ANSYS Analysis and Design Professionals\" – a platform dedicated to the analysis and design of pressure vessels. It has already attracted over 4,000 professionals in the field of stress analysis from various design institutes, engineering companies, manufacturing units, and universities across the country.