Towards a simplified assessment procedure for corroded nozzles in pressure vessels

Presented at ASME Pressure Vessels and Piping Conference, Boston, Massachusetts, USA, 1-5 August 1999.

Abstract

The integrity of pressure vessels can be compromised by corrosion in the nozzle region. There is therefore a need to assess the fitness for service of corroded nozzles using a simplified assessment procedure, even if this is only to ensure safe operation to the next shutdown. This review will examine alternative assessment procedures for the structural assessment of the integrity of corroded nozzles using various design methods such as the ASME area replacement rules (ASME VIII Division 1, 1995), the Welding Council experimental design rules (Rodabaugh, 1988) and the elastic stress rules utilised in BS5500 (BS 5500, 1997). Each method will be assessed against criterion such as safety and minimum tolerable wall thickness. The distance from the branch opening, at which simpler less conservative assessment methods can be used, will be discussed.

Introduction

Table 1 Summary of assessment techniques

1. the area replacement method of ASME VIII Division 1;
2. the Welding Council experimental design rules
3. the elastic stress analysis method used in BS 5500.
   

The first two methods have been adapted for use in the proposed draft recommended practice (API RP 579, 1997). The choice of which procedure to use for the assessment of corroded nozzles is not dependent on which standard was chosen to design the vessel. Instead the choice of assessment method is dependent on the input parameters available and which method will ensure the continued structural integrity of the vessel without being unnecessarily over conservative.

Attention is also drawn to the fact that whereas the original design codes provide comprehensive guidance on the avoidance of global collapse of the nozzle shell region, they provide little guidance on collapse of the local ligament beneath the corrosion defect. In ensuring the integrity of the corroded vessel, this mode of failure, together with other mechanisms such as fatigue, brittle fracture and buckling must also be considered.

Area Replacement Method

The area replacement rules assume that the material removed from the shell for the opening must be balanced by extra reinforcement material elsewhere. The principal concern of the area replacement method is to prevent global collapse of the shell near to the opening. Using the area replacement design rules for the assessment of corrosion near to the nozzle cannot, however, exclude the possibility of local collapse of the ligament under the corroded region.

It has been suggested that the area replacement rules may be reinterpreted so as to consider the area available for replacement in the nozzle wall as the original available nozzle area minus the defect area. This approach is essentially the same as that suggested in the draft recommended practice for the assessment of general metal loss in the draft API RP 579 (where the nozzle wall thickness is based on the average wall thickness). This gives a greater chance of the corroded nozzle being acceptable than where the nozzle is assessed on the basis of the minimum thickness. It should be noted however, that this approach appears to be (in some circumstances) considerably less conservative.

The Welding Research Council Experimental Design Rules

Elastic Stress Analysis Approach

The method has the advantage that it covers a much greater variety of nozzle designs than the method proposed for assessment of locally thinned areas in API RP 579 and has a firmer theoretical foundation than the area replacement model. In addition, it is claimed that the elastic stress method gives a more uniform factor of safety across a wider range of nozzle and shell geometries, than the area replacement method.

Discussion of Proposed Assessment Methods

• safety
• limits of nozzle extent
• consideration of external loads
• local ligament collapse
• comparison of calculated minimum wall thickness
• ease of use
  

Safety of Different Design Methods

It is not possible to give a measure of the absolute conservatism or safety of any of the design methods investigated in this study. However, all the design standards considered in this review have a history of successful use, based on operational experience. In the case of ASME VIII Division 1 this has been accumulated over a period of 100 years use (Holt, 1996).

The Welding Research Council rules are based on lower bound data from a number of destructive tests and can therefore be expected to give a conservative estimate of the design strength. However, it is not known how many pressure vessels have been designed using these rules and their operating performance is unknown. It should also be noted that the Welding Research Council design rules are not necessarily safe where the nozzle is subject to external loading.

The elastic stress rules described in BS 5500 have a history of 24 years of operational experience in the UK and overseas. The BS 5500 rules can therefore be considered as validated by safe custom and practice.

The use of any of the above design codes should prevent global collapse of the corroded nozzle. However, it is also recommended that steps are taken to prevent local collapse of the corroded ligament. This is done by specifying a minimum wall thickness.

In addition to global collapse and local ligament failure, consideration should also be given to other modes of failure. These may include brittle failure and fatigue and failure by buckling in the presence of compressive loads.

The Limits of Nozzle Extent

The 'die away' lengths specified in BS 5500 superficially imply that for corrosion outside this zone, other methods of assessment than those described in BS 5500 may be used to assess the structural integrity (e.g BS 7910, 1999). However, it should be noted that where external loads to the nozzle are considered (e.g. moment loading of nozzle) the 'die away' length may be greater.

The choice of whether to assess corrosion using the nozzle integrity model described in this review or the corroded pipe assessment procedure (such as ASME B31G or that described in BS 7910:1999) depends on whether the stress system at the corroded ligament is best described by the nozzle stresses or by the simple pipework stress relations. Accordingly, it was initially thought that the nozzle die-away length could be effectively taken as h =

along the vessel shell using Timoshenko's criterion (Timoshenko, 1959). However, independent studies of the effect of separation distance on the collapse pressure of a corroded nozzle done at TWI, show that the collapse pressure is much lower than one would normally expect for a flaw in a 'normal' equivalent pipeline, even when the separation distance is much greater than that suggested by the Timoshenko criterion. Clearly, the size of the effective nozzle region needs further research, both for flaws subject to internal pressure only and for flaws subject to internal pressure and external forces. However, a pragmatic (though probably over conservative) solution to the problem would be to consider the nozzle to end at the first flange and to consider all corrosion within this region to contribute to global collapse of the nozzle.

Consideration of External Loads

External loads are likely to lead to high discontinuity stresses resulting from the need for the shell and nozzle to maintain compatibility. Neither the area replacement rules nor the Welding Research Council method make any allowance for external loads on the nozzle, but from previous practice one would expect that a nozzle assessed using the area replacement rules is capable of carrying some external load.

The BS 5500 elastic stress method provides guidance about how to incorporate external loads into the calculation of an acceptable nozzle thickness. The elastic stress method should therefore give a reliable measure of the fitness-for-purpose of the nozzle when subjected to an external load provided the relevant maximum and minimum dimensions are used to assess the nozzle and that the external loads can be quantified or the pipework can be shown to have been designed to a suitable standard. The BS 5500 rules require that, where the nozzle is subject to external loads, the minimum thickness, regardless of the position of the corroded defect, should not be less than that given for the minimum branch thickness, allowing for mill tolerance. A typical manufacturer's tolerance is given as 12.5% of the wall thickness. Using this criterion has the advantage that it can be argued to be conservative through experience, though it is an empirical requirement.

Prevention of Local Ligament Collapse

Neither the area replacement rules, nor the elastic stress arguments of BS 5500 nor the Welding Research Council method were formulated with the aim of providing an assessment method for corrosion. It is proposed that they can be adopted to prevent global collapse of the nozzle, but this does not safeguard against the possibility of local collapse of the corroded ligament. In the design codes, the minimum thickness is generally stipulated as being sufficient to prevent the branch pipe failing. In many cases, the pressure in the vessel will be so low that this thickness is small in comparison with the minimum thickness required for external loading.

When these criteria are used with corroded vessels and based on the minimum thickness of the branch, they may be particularly conservative. This is particularly so when compared to the criteria for pipework. Four alternative criteria are postulated for the prevention of local collapse of the corroded ligament:

1. Limiting the minimum corroded wall thickness to that recommended in BS 5500 or ASME VIII, but making an additional allowance for mill tolerance.
   
   
2. Limiting the minimum wall thickness to 0.5 of the minimum design wall thickness or 3.2mm (as suggested in API RP 579 Level 1 rules for general metal loss). This criterion has the advantage that it is less conservative than the first and that the limit of 0.5 of the minimum design wall thickness partially avoids the problem of highly irregular corrosion disturbing the stress field at the crotch corner. However, safety cannot be guaranteed and the method conflicts with the requirements of BS 5500.
   
   
3. Conducting a limit load analysis on the corroded region. This is a complex analysis and requires information on peak stress, bending stresses and effect of interaction on the local ligament collapse stress. It is hampered by a lack of knowledge of the stress system.
   
   
4. Reinterpreting the ASME VIII and BS 5500 minimum branch thickness requirements by stipulating a limit to both the second moment of area and cross sectional area of the corroded branch in place of the current minimum thickness criterion. Use of these limits would greatly reduce the allowable minimum remaining thickness. However, more research would be needed to validate this interpretation.

The third method is the most rigorous but is unlikely to prove useful unless further research is done on the parameters affecting local collapse. The practical method to prevent ligament failure is to limit the corroded thickness to that recommended in the first option.

Comparison of Minimum Allowable Nozzle Wall Thickness using Different Design Standards

In order to compare the different standards, the minimum acceptable wall thickness for different nozzle-pressure vessel configurations has been calculated using the first method. Obviously it is impossible to consider all possible combinations of nozzle and pressure combinations. A brief survey of the minimum nozzle thickness for several representative nozzle sections is presented in Table 2. The individual cases were analysed using the Finglow software package (Finglow, 1996) which automates BS 5500 and ASME VIII design calculations. The BS 5500 elastic stress analysis method, ASME VIII Division 1 area replacement method and Welding Research Council method were used to assess typical nozzles. The minimum thickness required for a nozzle was calculated for twelve configurations comprising 100mm, 300mm and 500mm diameter nozzles in cylinders, thin hemispherical heads, thick hemispherical heads and 2:1 ellipsoidal heads. It can be seen that the minimum thickness for a nozzle in a cylindrical shell or 2:1 ellipsoidal head, as calculated using the BS 5500 3.5.4 method, is much less than the minimum thickness calculated using the area replacement rules of ASME VIII Division 1. However, there is very little difference between the minimum required thickness for a nozzle in a hemispherical head calculated using either the BS 5500 method or the area replacement method. Indeed, where the nozzle diameter was very large, the area replacement method actually resulted in a thinner minimum required thickness compared to the BS 5500 3.5.4 method.

The minimum thickness derived from the Welding Research Council method was also compared with the minimum thickness derived from the BS 5500 elastic stress method for nozzles in cylindrical shells. Table 2 shows that for two of the three cases examined, the minimum nozzle thickness obtained by the Welding Research Council method was greater than that recommended for the BS 5500 elastic stress method, while for one of the cases the minimum wall thickness was less than that recommended in BS 5500.

The minimum wall thickness for a nozzle calculated from BS 5500 could be further reduced by using actual yield strength data instead of specified minimum values.

In summary, it can be said that in many cases, the elastic stress analysis rules of BS 5500 result in a thinner minimum acceptable nozzle wall thickness than the ASME VIII Division 1 rules. However, it should not be assumed that using the elastic stress analysis rules always lead to a thinner acceptable nozzle wall thickness.

Ease of Use and Input Information Required

Superficially, it would appear that the BS 5500 3.5.4 nozzle design rules are complex to use, but both the design codes for the ASME VIII Division 1 guidelines and the BS 5500 elastic stress nozzle design rules can be implemented using commercially available software packages (e.g. Finglow). As part of the investigation, the author made a subjective assessment of the user friendliness of one software package (Finglow). In the author's opinion, the package can be learnt quickly and is easy to operate. There is no recognisable difference between the ease of using the BS 5500 elastic stress nozzle design rules or the ASME VIII Division 1 area replacement nozzle design rules.

A computer package could be easily developed to assess the Welding Research Council equations. However, it should be noted that these equations are very limited in their application.

Inspection and Maintenance

• How is the nozzle thickness to be averaged?
• How can the risk of brittle fracture and fatigue damage be minimised?
• What is the maximum acceptable inspection interval?
  

The inspection of vessels is addressed in API RP 510 (1992) and recommendations are suggested in the draft API RP 579. The prevention of brittle fracture is addressed in API RP 920 (1992) and by BS 7910:1999. It is recommended that pressure vessels and nozzles should be inspected in accordance with API RP 510. The integrity of the vessel can be assessed on the basis of minimum wall thickness within the nozzle region. Using the minimum wall thickness is thought to be potentially over conservative, however, although the averaging the nozzle thickness method may in practice give an adequate safety margin for safe use, there is at present insufficient evidence to recommend the use of thickness averaging for assessment purposes. Inspection should be capable of determining the minimum remaining wall thickness and detecting the presence of crack-like defects. Where crack-like indications are found, any of the above assessment procedure will have to be complemented by fracture mechanics calculations such as in BS 7910:1999.

The inspection of welds is particularly important. Where welds are corroded, these need to be inspected for crack-like indications and the corroded weld dimensions checked against the recommended weld dimensions for the nozzle. Ideally, nozzles should be inspected internally. Where this is not possible, BS 3923:Part 1 provides a procedure for the inspection for nozzle welds using ultrasonic inspection. This may need to be adapted to the inspection of corroded pressure vessels.

Local blend grinding is recommended where the nozzle is badly pitted. Blend grinding will reduce the risk of brittle fracture by removing local stress concentrations and reduce the rate of pitting corrosion.

Summary

In the absence of the risk of failure due to fatigue loading, brittle and ductile fracture, or buckling, two failure modes are recognised when a nozzle becomes corroded. These are global collapse of the nozzle region and collapse of the remaining ligament in the corroded region. A simple approach to the assessment of global collapse is to assume that the nozzle wall or pressure vessel thickness is equal to the minimum remaining wall thickness. The current codes give methods to evaluate the minimum thickness required for a nozzle in a pressure vessel based on an assumption of uniform nozzle and uniform pressure vessel wall thickness. This review has identified three fundamentally different approaches to nozzle design; the area replacement method based on ASME VIII Division 1, the elastic stress analysis method based on BS 5500 and the Welding Research Council method. These methods have been shown to be reliable through their use in the design of nozzles in serving pressure vessels over many years.

The elastic stress analysis method has a foundation based in theoretical stress analysis and, in many cases, avoids the over conservatism associated with the area replacement method. Additional advantage may also be taken of increases in the design strength resulting from actual material yield and tensile strength values compared to the nominal design strength based on specified minimum yield and tensile strengths used in the design of the vessels.

While the risk of global collapse can be minimised by adopting the elastic stress methods of BS 5500, comparatively little information is available concerning local collapse of the corroded ligament. In view of the uncertainty about the validity of any method for assessing local collapse, it is recommended that one adopts the current (conservative) practice recommended in BS 5500 for the assessment of corroded nozzles. There is considerable scope for reducing the minimum thickness below the BS 5500 recommended minimum limits. However, validation studies, possibly involving FEA, are recommended before any alternative criterion is adopted.

The elastic stress analysis procedure to assess corrosion at nozzles is aided by the use of purpose-written software packages and can be applied readily in practice. It is understood that a procedure based closely on the recommendations of this review has been adopted by a major oil company and has been applied safely to corroded nozzles. A procedure based on the recommendations of this review is illustrated in Fig.1.

Conclusions

Several pressure vessel nozzle design codes and assessment procedures for corroded nozzles have been critically reviewed. On the basis of the literature review it was judged that BS 5500 generally gave the best basis for a corrosion assessment. The following conclusions have been drawn:

• Global collapse of corroded nozzles can be avoided by assessing the nozzle using any of the following methods: the ASME VIII Division 1 area replacement rules; the Welding Research Council's method or the BS 5500 elastic stress design rules. The use of these methods to assess corrosion can be based on the minimum wall thickness.
  
  
• Comparison of the minimum wall thickness obtained using the area replacement, Welding Research Council or elastic stress methods of nozzle design, showed that in many cases the BS 5500 elastic stress method gave a lower predicted minimum wall thickness than the others. However, no method gave consistently lower allowable minimum wall thicknesses compared to the others.
  
  
• Use of a global collapse assessment criteria alone is insufficient to ensure fitness-for-purpose. Local collapse of the corroded ligament must also be considered.
  
  
• At present, there is insufficient evidence to assess the distance a flaw has to be from an opening, so that the collapse pressure is unaffected by the presence of the opening. This requires further research.
  
  
• The effect of external loads on the global collapse of the nozzle can be assessed on the basis of the guidelines given in BS 5500. However, the influence of external loads on the 'die away' length and local collapse of the ligamentare unknown.

The following procedure is recommended:

• The global collapse condition of the nozzle should be assessed using the BS 5500 elastic stress method. The remaining thickness of the nozzle should be assessed on the basis of the minimum remaining wall thickness.
  
  
• The collapse load of the remaining ligament should be assessed using the guidelines of BS 5500, but reducing the minimum thickness to 7/8th of that quoted in Table 3.5.2 of BS 5500 (1997).
  
  
• Under conditions of negligible external load, corrosion outside the 'die away' zone should be assessed using the guidelines for corrosion assessment of pipework.
  
  
• Where the procedure recommended in this review leads to the corroded nozzle being evaluated as unfit-for-service, either finite element analysis should be used to reassess the nozzle, or the pressure of the nozzle should bedownrated to comply with the requirements of BS 5500.
  

Further Work

• It is strongly recommended that the procedure advocated in this report is validated against finite element studies.
  
  
• It is recommended that ligament collapse, the effect of thickness averaging on conservatism, external loads, and the 'die away' length are investigated further.
  

Acknowledgements

The work was carried out within the Core Research Programme of TWI, which is funded by the Industrial Membership of TWI. Particular thanks go to Phillips Petroleum Company Norway for additional funding for this work.

References

Copyright by TWI, 1999