Browsing by Author "Beyers, Wilhelm Andre"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Evaluation and modification of air-cooled heat exchanger header box design procedure
    (Stellenbosch : Stellenbosch University, 2014-12) Beyers, Wilhelm Andre; Venter, Gerhard; Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.
    ENGLISH ABSTRACT: The purpose of the heat exchanger header box is to act as a high pressure manifold that redistributes process fluid from a central source to a bundle of finned tubeswhere it can be cooled. South African law requires that pressure vessels like this one must to be designed according to a pressure vessel design code, such as the American Society of Mechanical Engineers’ (ASME) pressure vessel design code. The two most commonly used header box designs are a plug type and a cover type header box. For the design of a plug type header box, ASME Section VIII Division 1 provides a full complement of rules and formulae necessary for producing a satisfactory pressure vessel design. Such a set of rules and formulae are however not available for cover type header boxes. To overcome this problem, industrial manufactures have developed in-house design codes focussing specifically on the design of cover type header boxes. These in-house codes draw in part on existing formulae in the ASME code, but rely primarily on calculating the strength of a header box using a simplified model consisting of simply supported beams, each simulating a different part of the header box. Recently there has been some concern regarding the accuracy and validity of these in-house design codes. This project has sought to address some of these concerns by evaluating one such in-house design code and comparing its results with those from finite element analyses. For the purpose of this comparison it is shown that the header box’s structure can be simplified to a representative 2D finite element model while still yielding trustworthy results. The results from the comparisons with these 2D models showed that the in-house methods produced non-conservative estimations for certain stresses in a header box. However, after analysing multiple sample designs, it was also found that header boxes that had been designed using this method were all substantially over-designed. Two major contributing factors that lead to this over design were identified. The first was thatmaterial and manufacturing constraints frequently necessitate that material thicknesses be increased. The second factor was that because of the concern linked to the validity of these in-house design methods, designers are inclined to further increase the material thicknesses to account for any uncertainty. To address this problem, a new 2D finite element software package was developed, on an open source platform, to accurately analyse the structure of cover type header boxes. The software is designed to integrate directly with the platformcurrently being used to performthe calculations for the in-house code and includes an optimiser that takes the material and manufacturing constraints into account. The software can be used to validate any existing designs, as well as providing optimal designs and accurate stress predictions for new header boxes, thus reducing uncertainty in the design process.
  • Thumbnail Image
    Item
    Predicting structural behaviour of pressure vessels using large scale meta-modelling applied to plug type heat exchanger header boxes
    (Stellenbosch : Stellenbosch University, 2017-12) Beyers, Wilhelm Andre; Venter, Gerhard; Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.
    ENGLISH ABSTRACT: Plug type header boxes are predominantly designed according to the American Society of Mechanical Engineers’ Boiler and Pressure Vessel Code. The two design methods most often employed from this code are ‘Design by Rule’ from Division 1 of the code and ‘Design by Analysis’ from Division 2 Part 5. While ‘Design by Rule’ is relatively simple to implement and relies only on a set of hand calculations which produce immediate results, it is widely known to be conservative. ‘Design by Analysis’ on the other hand relies on finite element analysis, which yields more accurate results, but is more costly and time consuming. In this study a new analysis tool was developed in the form of a metamodel, based on finite element results, which predicts the structural behaviour for various plug type header boxes. The purpose of this tool is to provide a designer with real time predictions of the stresses in a header box, as is the case for ‘Design by Rule’, but with accuracy similar to a finite element analysis. In order to achieve this goal, a software tool set was developed which automates the process of setting up, simulating and post-processing the results of a finite element analysis. This made it possible to generate numerical results on a large scale, in order to collect enough data to train an accurate meta-model. Using this tool set, a number of less complex meta-models were initially created to test the approach and refine the procedure employed. These tests were performed using very large training sets and showed that high (4th and 5th) order response surface models were required to accurately approximate the structural behaviour of plug type header boxes. The challenges associated with fitting such high order response surfaces were addressed and these models were systematically scaled up in complexity until the final meta-model was constructed. The final meta-model achieved the desired goals of providing accurate stress results, in real time, for a plug type header box. Designers can use this model to search for optimal designs and identify what the structural effects are when individual header box dimensions are changed. This will allow for detailed insight to be gained of the structural behaviour of plug type header boxes in a manner which has been unavailable in the past. The implications of such knowledge will expand the field of knowledge surrounding these pressure vessels and open the door for the implementation of possible new design methods in the future.