Quick Search

684 Remoteness of contacts and lack of supervision

Report ID: 684

Published: Newsletter No 53 - January 2019

Report Overview

A correspondent has been thinking about the problem of lack of supervision in construction generally and has come to the conclusion that this is all part of a wider problem, namely the general remoteness of many things.

Report Content

A correspondent has been thinking about the problem of lack of supervision in construction generally and has come to the conclusion that this is all part of a wider problem, namely the general remoteness of many things; a lack of police on the beat, switchboards to take calls, tickets from railway stations, no operators’ manuals for complicated tools and equipment, everything being on-line and so on. In the correspondent’s view, the Professional Institutions are going the same way.

We are missing something, according to the reporter. Maybe it is trying to automate things too much as they feared in the 1930s, but the reporter thinks it is slightly worse than that. Those responsible for managing many things these days prefer to see some remote camera or other sensor and a central spreadsheet of results. The correspondent has come across this frequently with attitudes to structural monitoring; the order of the day seems to prefer sophisticated remote sensors periodically feeding back data to some central control a long way away.

Experienced eyes in the field, and immediate warnings are much more important. That is quite an obvious example of general remoteness but there are more subtle ones in construction work due to the way civil and structural engineering things are organised.

Intelligent computer systems have not really progressed beyond the Pavlovian illusion of intelligence, and too much faith is being placed in them, and promised smart solutions. What brought this to mind recently was that the correspondent was on a panel interviewing a candidate for professional membership and the question of verifying complicated calculations and adjustments to field measurements associated with a massive railway construction project came up.

It turned out that the approving body for the construction of the project only ever re-ran computer analyses of the contractor’s own field observations - no mention was made of checking the field observations themselves (the fundamental things which might be wrong). What use therefore is this chain of 'approval' with a possible weak link? We have in that a clear case of a large organisation refraining from getting its hands dirty when it comes to engineering.

Comments

This is an interesting report. No one should doubt that behind every failure there is generally a human action. The management of safety should assume someone will make a mistake, so all work should be checked and verified: that is office work and site installation.

The check should be to verify that the ‘model’ is appropriate, that the input data is appropriate, and that the output makes sense. The checker should think hard about what has been omitted or overlooked.

However, modern technology offers huge advantages in structural monitoring that did not exist before. So, we need to assess how to make best use of them: the reality will be there just isn’t the manpower to monitor everything. But there is no point in collecting data if it is not processed or acted upon. In 1994 the collapse of the Heathrow tunnel was partly due to no one acting on the mass of (ground movement) data collected and which was an essential component of NATM tunnelling methodology.

It is worth mentioning the Reason’s Plates (Swiss Cheese) model in terms of management. The point is that no system can be allowed to have a single point of failure (single slice of Swiss Cheese). Also, multiple ‘slices of cheese’ need true independence otherwise the holes will align and two is no better than one. Reason’s analogy can be applied to management systems just as much as to safety control systems.

Alerts from remote monitoring should trigger discussion among the engineers involved about the likely causes. Competent engineers must interpret remote collected data, at least until there have been much greater improvements in artificial intelligence. Whilst the reporter thinks that the checker should have independently checked the site data, it is unlikely that was ever in their scope. Maybe the more fundamental question is how major projects are set up such that the responsibilities for design, construction and checking are robust and reliable.

 

CROSS depends on you for reports - if you have experienced a safety issue that you can share with CROSS, please Submit a CROSS Report which is treated as confidential

If you have any comments regarding this CROSS report, please Submit Feedback

View other CROSS reports published in Newsletter No. 53


Email Updates

How to Report

Online submission:
Submit by post: