Civil Engineering thin cylindrical shells such as silos and tanks are normally subjected to axial compression that arises from a stored solid, wind, earthquake, self-weight or roof loads. The walls of these shells are very thin, generally of the order of 6 to 25 mm, and massively less than the radius, which is typically 5 to 30 m. They are thus very thin shell structures, like those of rockets, spacecraft, motor vehicles and aircraft. The commonest failure mode is elastic buckling under axial compression. It has long been known that the buckling strength of a thin cylindrical shell under axial compression is very sensitive to tiny deviations of geometry, reducing the buckling strength to perhaps 10 or 20% of the value for the perfect structure. A normal internal pressure usually accompanies the axial compression, caused by stored granular solids or fluids. At relatively low pressures, the elastic buckling strength under axial compression rises, but an elastic-plastic buckling phenomenon intervenes at higher pressures, causing a dramatic decrease in buckling resistance associated with an elephant’s foot collapse mode. To construct such large shells, the fabrication technique is generally the assembly of many rolled plates or panels, joined by short longitudinal welds and continuous circumferential welds. The process of welding produces a distinctive geometric imperfection form at each weld joint, which in turn is extremely detrimental to the shell axial buckling carrying capacity. The strength may be further reduced by slight misalignments between adjacent panels, or in bolted construction, by vertical and horizontal lap splices. Due to the pattern of loading, both the axial compression and internal pressure increase progressively down the wall. Accordingly, practical construction usually uses a stepped wall, formed from panels of uniform thickness, but with larger thicknesses at lower levels. Since the loading varies smoothly, but each panel has constant thickness, the critical location for buckling lies at the base of a panel. But the greater thickness of the lower panel can usefully enhance the buckling strength of the critical panel above it. This thesis presents an extensive computational study that examines all the above influences, divided into chapters that are outlined here. A full exploration of the effect of the cylinder length on the perfect and imperfect elastic buckling strength is presented in Chapter 3. In Chapter 4, the elastic-plastic buckling resistance of imperfect cylinders is described, including strain hardening. These lead to many capacity curves, for which the key parameters are extracted. The effect of internal pressure on the buckling resistance of imperfect elastic cylinders is explored in Chapter 5. Chapter 6 studies the effect of high pressures that produce elastic-plastic elephant’s foot buckling at circumferential welds in imperfect shells. Next, a step change in plate thickness is studied in Chapter 7 for imperfect butt jointed cylinders with and without the internal pressure. Chapter 8 presents an exploration of the effect of plate misalignments at a circumferential joint, as well as the full misalignment of a circumferential lap joint in bolted construction. These are investigated in both the elastic and elastic-plastic domains. The entire thesis is conceived in the context of EN 1993-1-6 (2007) and the ECCS Recommendations on Shell Buckling (2008). This research has shown significant weaknesses in both the concepts and the detailed rules of these standards. Many conditions are found where either the standard is unnecessarily conservative, or its safety margin may be too low. Thus, some new provisions are proposed for each of the above practical problems. These are expected to provide useful knowledge for the design of such structures against buckling in the future.

Shell boilers used in Sweden and Europe has to be certified before being used. Kiwa Sweden performs the role as an accredited third party for control services within most industries in Sweden. Kiwa Sweden has acquired an analysis program called  sPEeD12953 from the danish company Danish Exergy Technology (DXT). The analysis program sPEeD has previously not yet been validated and therefore, Kiwa can not use this program as a stand alone tool in their reviews of shell boilers. In order for Kiwa to be able to use the analysis program as more than just a support program, it has to be validated and quality assured. There are two versions of the program sPEeD, one called 2002 version and one called 2016 version. In this thesis, only the version called 2016 is investigated. This thesis was aimed to validate the analysis program sPEeD12953. The result of this study shows that the program sPEeD12953 performs well on most parts and gives good results. However, some of the calculation part of the program gives the wrong result according to the standard. Most of these errors are on the conservative side. Conservative errors leads to that if the user obtains a result that shows that the part of the boiler is okay, then it can be used. However, if the user obtains a not okay result in sPEeD, the user is required to perform additional calculations with another tool. A few results ended up with non-conservative results and these parts of the program can therefore not be used without the use of additional tools. The program can because of this not be fully validated and the software provider has to resolve these issues before the software can be considered fully validated. / Eldrörspannor som används i Sverige och Europa måste certifieras innan de kan tas i bruk. Kiwa Sverige är ett ackrediterat tredjeparts organ inom de flesta branscher i Sverige. Kiwa Sverige har köpt in ett analysprogram som heter sPEeD12953 från det danska företaget Danish Exergy Technology (DXT). Analysprogrammet sPEeD är ännu inte validerat och därför kan Kiwa inte använda detta program som enda verktyg i sina utvärderingar av eldrörspannor. För att Kiwa ska kunna använda analysprogram som mer än bara ett stödprogram, måste det valideras och kvalitetssäkras. Det finns två versioner av programmet sPEeD, en som heter ”2002 version” och en som heter ”2016 version”. I detta examensarbete är det bara den variant som heter 2016 som undersöks. Detta examensarbete syftade till att validera analysprogrammet sPEeD12953. Resultatet av denna studie visar att programmet sPEeD12953 presterar bra på de flesta delar och ger bra resultat. En del av programmet ger dock fel resultat enligt standarden. De flesta av dessa fel är konservativa. Konservativa fel leder till att om användaren får ett resultat som visar att den del av pannan är okej, kan den användas. Men om användaren får ett inte okej resultat i SPEeD, måste användaren utföra ytterligare beräkningar med andra verktyg. Däremot visade några resultat icke-konservativa resultat och dessa delar av programmet kan därför inte användas utan andra beräkningsverktyg. Programmet kan på grund av detta inte klassas som helt validerat och programleverantören måste justera och komplettera programmet innan det kan anses fullt ut kvalitetssäkrat.