Holdfast Training Services Trial Blog

I mentioned in my first blog that the construction of the ground floor slab is considered a critical milestone. With that for context you would assume that everything possible is being done to ensure the concrete pours on this slab are completed on programme. This brings me nicely on to the topic of this blog.

Early last week, only 60 mins before a significant concrete pour (50m³) was ready to begin on a key part of this slab, an issue was identified during final reinforcement inspections. This caused a fair amount of aggravation and stress on site between the consultant engineers and concrete package sub-contractors .

The issue relates to the column in Fig 1.1 and in particular the depth of fillet welds between the web and flanges. The column protrudes through the GF slab and was scheduled to be encased in concrete up to GF level as part of this pour. The engineer consultant, during his final check of the slab reinforcement bar, identified an issue with the welds on the column that stopped work until the senior structural engineers at the design office could be consulted.

Fig 1.1 – Incorrect Weld Configuration on Column L8

As I know the audience reading this blog adore explosives, and in particular blowing up military bridges on exercise, I thought they might be interested to hear that many of the structural elements have been designed to resist an explosive blast detonated at very close proximity.

Robustness in a building is usually achieved through one of a number of approaches. The most common is to design alternative load paths in case a structural element fails. On this project however, due to a late notice client dictated design change, this was not a viable approach. Robustness is therefore only provided through the alternative ‘protected element’ method. By this I mean key structural elements are designed with adequate individual robustness and additional protective measures (i.e. encased in sacrificial concrete or steel) to ensure they do not fail even when exposed to the designed worst case load condition.

In this particular column the weld design has been specified for this worst case condition. Unfortunately the column was not constructed in accordance with the design drawings. To provide suitable shear resistance in this element, the Blast Engineering Consultants specified a 70mm multi run fillet weld along the full length of the column to the base of the ground floor slab. As Fig 1.1 shows, the steel contractors only installed the fillet weld from the top of the column to the first web stiffener, leaving a considerable length of this column with minimal 15mm deep fillet welds. To try and understand the magnitude and context of the site engineers concern I tried to conduct some very basic analysis of this section.

In the first instance I modelled this problem as simply as possible and effectively considered the column as an I beam bending in one axis with an 8.6MN worst case vertical shear reaction force. To be clear from the outset, it will become apparent later in the blog that this model is far to simple for the complexities encountered in this problem.

Students on the civil stream will fondly remember the “Say It” equation which allows calculation of shear stress along a particular plane at any point in a bending beam. One of the key components in this formula is the thickness (b) of material providing longitudinal shear resistance along the assessed shear plane. The greater the (b) the smaller the shear stress. In steel plate sections a fillet weld exists to simply transfer the shear stresses from the flanges to the web so that the top and bottom flange can work together as a single beam (ie they know about each other). It’s  worth noting that because these columns are steel plate girders, there is no monolithic connection between the web and flanges to provide an additional contribution towards this shear resistance. I have simplified the diagram to sum this up and illustrate why the reduction of this weld from 70mm to 15mm might be a problem.

Fig 1.2 – Simplified Column Section Model

Diagram A shows the steel column, as installed, below the web stiffener. Along this length the fillet weld installed is only 15mm in width each side of the web, this provides a total (b) of only 30mm. Diagram B shows the steel column where the fillet weld has been correctly installed. This weld, 70mm on each side of the web, provides a total (b) of 140mm of resistance along this shear plane.

In accordance with BSEN 93-1-8 Para 4.5.2 you actually need to use the effective Throat thickness (a) of each weld in this calculation (This provides a conservative estimate) which can be acquired with some simple trigonometry. I calculated weld throat values of 10.6mm and 49.5mm for case A and B respectively. These figures are then doubled through most of the calculations because we have a weld on both sides of the web.

I then plugged these dimensions and the known constants into the “Say It” equation, with the two different welds considered, to analyse the varying shear stresses induced as a result of the designed maximum blast load occurring on the column:

Calculation Details:

Max Vertical Shear Force on Section Vz = 8.6 MN

Second Moment of Area = 32.77 x 10⁴ mm⁴

Distance from NA to Centroid of Area above shear plane = 200mm

Area above Shear Plane = 38400mm²

Width of Material under shear (Weld Throat): Case A = 21.2mm

Width of Material under shear (Weld Throat) : Case B = 99mm

After I’d run the calcs through using the information  above I got the following shear stress values through the welds in each case:

Shear Stress in Weld:

Case A: 950.7 N/mm²

Case B: 203.6 N/mm²

I also calculated the greatest shear stress likely to be seen in the I section for both examples in order to allow further comparison. We know this appears along the NA where the web in this case is 60mm. I ended up with shear stress values as below.

Shear Stress along NA – Web:

Case A: 371.1 N/mm²

Case B: 398.78 N/mm²

When you look at this logically, it raises a few interesting points. In Case A the shear stress expected in the weld is over 2.5 x greater than that seen in the web on the NA. Therefore, when affected by the blast load, the weld is likely to fail long before the web of this section.

In Case B however the weld sees an expected stress that is approximately half the shear stress seen along the NA in the web. In this case it is likely the web would fail before the weld. This suggests the 70mm weld is therefore larger than actually required. Generally a design throat thickness of an I section, doubled to account for both sides of the weld, should be a similar depth as the web thickness on that section. Any additional weld depth is most often unnecessary because the failure risk is transferred to the web.

Using BSEN I then checked the actual permissible shear stress in each weld. In both cases I calculated this to be 230.9 N/mm². When you compare this to the expected stresses, it is clear that the fillet weld in case B is capable of carrying the design stress of 203.6 N/mm² but not the 950 N/mm² of case A; clear indication that the 15mm weld is grossly under strength for the worst case load condition assumed in this model.

I mentioned that my first model kept this analysis as simple as possible. In reality the problem is considerably more complex for a number of reasons. Firstly, I assumed the blast occurred directly perpendicular to top flange. Blasts occurring at different angles to the section would undoubtedly affect the results. In addition to the 8.6MN vertical shear reaction load I considered on the section there is also a 9.5MN axial load from the weight of the building applying a compressive pre stress through the column. Furthermore, the complex nature of the dynamic/impulsive blast load on the section is also far more challenging to categorise; it is affected by numerous variables including charge size and type, stand off, ambient pressure and charge proximity to the ground, amongst others. The response of a steel column to a blast load is also influenced by stiffness, its dimensions, vibration period and strain-rate. The point being, this is far too complex to analyse with such a simple model.

What is very clear is that no one in the project team seems to understand the science or engineering behind the blast design analysis. All we know is that there is a specified design explosion and the consultants tell us the size of elements required to provide robustness. If I can learn about the science and engineering models applied in the space and time between the immediate explosion and its impact on the column I will probably know more than most of my immediate colleagues and have a potentially interesting thesis topic. Fortunately I have managed to arrange a visit to the blast design consultants test facility in the coming weeks to see first hand how they model their problems to produce the design output. With any luck they will also let me blow up some steel columns in the name of thesis or TMR research.

In practical management terms the project team on site got irritated by this problem because the column had been in position for three weeks before the issue was spotted. Had we employed more thorough quality assurance inspections, it is likely this issue would have been picked up earlier.

This incident is also a potential indication of a more commonplace issue. From a brief investigation I am certain that the column was inspected and signed off imediately after installation. The likely problem therefore, I’m speculating only, is that the individual who inspected and signed off the steel either didn’t do it thoroughly, made a mistake or potentially didn’t know the detail of what they were looking for when completing this process. Which raises the question of how to best ensure QA inspections of completed work are only conducted by individuals capable of understanding the technical importance of the details contained on drawings and specifications.

Once I have got more information on the modelling methods used by the blast consultants I will attempt to publish another blog on the subject and its affect on my own analysis of the issue identified.