Holdfast Training Services Trial Blog

Evening all,

A short and sweet post.  As a lowly E&M engineer in training I am interested in some second opinions on the deflection of steel structures under load.

This week I have been occupied with installing temporary tunnel ventilation fans in the -5 level of Bond Street Station.  The fans weigh around 3500kgs and will be in operation for around 18 months.  The first of four fans was due to be moved into place today using a porta-gantry onto a simple steel frame.

The fan being lowered into position on the support frame

Upon lowering the fan onto the support one of the operatives spotted a part of the frame deflecting under the applied load.  The web of a channel section was bending inwards as the chain hoist lowered the fan, distorting the anti-vibration mounts.

Apologies for the poor photo. If you draw a line along the top and bottom flanges of the C-section they are no longer parallel; causing distortion to the anti-vibration mounts.  The lift was stopped before the full load was applied.

This would all be highly interesting except for the fact that I am only SMSTS qualified person on site and conducted the brief for the works.  It also came down to me and a grad mechanical engineer to judge whether the deflection the channel was showing was acceptable, and whether to continue lowering the full load onto the deflecting frame.  After some teeth sucking and discussion with the lift AP we decided to lift it back off and question the frame designers.  Queue a lot of Northern blokes pissed off that they had wasted a days’ work.  Reminds me of being a Troop Commander.

On removing the load the channel returned to its original shape with no permanent deformation.

Was it the right decision to stop the lift?  Or can you expect visible deformation on a channel loaded in this direction?  Is this the right steel section to use for this application?

One of my recommendations in an early AER was that the APM method (bottom-up) of estimating was useable and effective in forecasting my budgets.  Another was that line items in a risk register should not be expected to occur in isolation – they often work in alliance e.g. groundwater slowing shaft excavation and affecting tunnelling rate (critical path). Three separate cost codes which were all affected by one item.  In my next breath I am going to retract my first recommendation and offer you risk-based cost estimating.

My job in design is a small one worth less than my annual salary, but there have been a few golden nuggets to take away.  One of my deliverables in the detailed design for a highway widening scheme is an estimate on what the works are likely to cost, using the Client’s risk-based cost estimate template. It is essentially a Monte-Carlo analysis (M-C).

Similar to the RMS (Root Mean Square) method, the M-C  identifies that the likelihood of all maximum risk values occurring on one project is low.  Using data which the user inputs, it models thousands of possible scenarios (or risks) on a project occurring to greater and lesser degrees.

I modelled a simple fantasy project with the same worst, most likely and best case costs using the PERT analysis, and compared the results with the M-C. The end product is a normal distribution curve (screenshot below) which gives you the likelihood or confidence (%) of the project costing ‘X’ (£).

The PERT results gave me a 50% chance that the project would cost $232k, and a 95% chance that it would cost $268k.

The M-C 50% result was also $232k, but I could be 95% sure it could be done for $249k.

So what? As a Client with numerous projects in a programme, you’d be better informed on where to put your money.

I’ve now used bottom-up estimating, PERT, RMS and M-C. All have their pros and cons but the RMS and M-C must be considered the better options. I would argue that unless you have the M-C software package, or Damo to build one on excel, the RMS is sufficient.  Both share the same limitation in that they inevitably spit out numbers based on subjective information – garbage in is garbage out. There lies the risk within the risk analysis.