Review of

R.A.S. Siriwardane Arachchilage, M. Poljansek, S.P.M. Eekelen, Bend And Break Shell Structures Design exercise, report for course CT3280 Shell Roofs, Delft University of Technology, Department of Civil Engineering and Geosciences, 2023, online: phoogenboom.nl/B&B_schaal_report_10.pdf

by dr.ir. P.C.J. Hoogenboom, Delft, 3 January 2024

 1. Complete report. Excellent report structure. Few typing errors. Handed in on time (22 Dec 2023).

 2. Include "Delft University" and "Department of Civil Engineering and Geosciences" on the title pages.

 3. A preface could have been added to explain the context of the work (minor, 4 weeks, 5 credits, ...). This is useful for a reader who finds the report 5 years from now.

 4. Good summery

 5. On page 4 the unit is wrong. Stress is measured in kN/m^2. Moreover, the compressive strength of pine is about 36 N/mm^2. This is 1000 times larger than 36 kN/m^2.

 6. The construction process is described clearly. I understand that at the start of the course, you needed more laths of more dimensions. Thanks for this feedback. Good to know for next time.

 7. The technical drawings with dimensions are clear.

 8. I think this design has much torsion in the hoop beams. Timber is not good in carrying torsion. The structure as built has much less torsion in the hoop beams.

 9. There should also be a drawing of the structure as built.

10. It was decided to load the main arches only. How is the snow load transferred to these arches? There should be roof cover, insulation, plates, beams and arches. These are not part of the model and not tested for strength. This should have been mentioned in the report.

11. The load distribution system is not statically determined. So, the stiffness of the ropes determines how much load goes to the top and how much goes to the arches. The load should have been distributed as on the real size structure (selfweight and snow).

12. Good that there are different chapters on "Test set-up" and on "Results/Discussion".

13. Does the computer model represent the model roof or the real roof? (No difference in stresses but a large difference in deflection.)

14. The Von Mises stress is accurate for yielding of metals. It is not for wood. The other stresses that were used are accurate.

15. The largest shear stress occurs in singularities. This stress does not occur in the physical model. It should have been ignored.

16. On page 14 is written "... stress did not linear increase ...". You think that SCIA did a non-linear analysis? Are you sure? I think SCIA did a linear analyses here. Nonetheless, if SCIA did non-linear analysis here, the arguments and the followed procedure make sense.

17. Op page 17 a knockdown factor of 1/6 is applied to the nonlinear analysis result. This is wrong. The knockdown factor should be applied to the linear buckling result.

18. The computer model is realistic. Each of the three types of analysis has been performed. Most of the results have been interpreted correctly, which is not an easy thing to do.

19. Good photo of loading the test set-up

20. Good photo of the collapsed structure

21. On page 23 is written "Thirdly, ... curving of an element allows redistribution ...". This argument makes no sense to me. The curvature is included in the computer model. It is not a source of extra strength. 

22. Except for the previous comment, the evaluation of the test is very solid. My complements.

23. On page 25; how is 15.7 kN/m2 obtained? Calculation results must be reproducable. My calculation: 3.24 kN / (0.85 m x 0.85 m) = 4.5 kN/m2.

24. The conclusion is positive and clear, which is good. However, can this shell be built 10 times as large? This is not answered. In fact, a roof spanning 8.5 m that can carry 4.5 kN/m2 (selfweight and snow) is quite safe.