Heathrow T5 Baggage Handling System - the jury is still out

June 18, 2008

There are some things you just can't test for

Posted by: Dave Evans

I said in an earlier post that I was looking forward to hearing Huub van der Wouden, the test manager for the Heathrow Terminal 5 Baggage Handling system, speak at the BCS SIGIST, and I think his presence had a lot to do with the unusually high attendance at the conference today.

What looked like a scoop (and an act of bravery) turns out to have been an extraordinarily good piece of anticipation on the part of the SIGIST organisers, as the speaking slot was arranged long before the fateful T5 opening day. Huub, of Dutch consultancy Transfer Solutions, raised a laugh when he opened the talk as he described how his feeling of honour at being invited to present at SIGIST turned to horror after the events of March 27th unfolded. This explained the rather uncontroversial and upbeat description of the presentation in the SIGIST programme.

Huub and Roger Derksen (from Vanderlande, the company contracted by BAA to build the system) gave engaging presentations about the testing carried out, showing a lot of interesting test techniques and a good description of the whole process.

They went through the strategy of starting testing as early as possible, working within the logistical constraints of the construction process. There were videos of models and simulators, and of 'real-world' load tests costing thousands of pounds per run. We heard of the highly layered architecture, of factory testing and site testing, ideal bags and real bags, of the testing of messages and moving parts. It was all good.

The audience may have expected this to lead up to a dramatic climax, where we found out where it all went disastrously wrong on opening day, but it was not to be. There was a discernible collective sagging of shoulders as Huub explained that he and Roger were under orders not to talk about "that side of things", and probably would not be able to do so for at least twelve months. If there was a smoking gun, they were not going to show it.

As we headed for the pub after the event, I asked Huub to put me out of my misery. "Was testing to blame for failing to anticipate the problems? Or did testing reveal problems that were ignored by management?" He gave the wry smile of a man who accepted that despite the cordial welcome and the disclaimer, the day would not end without some prickly questions.

He refused to be drawn on details of course, but hinted that the problems arose from human and logistical concerns, most of which occupy the grey area between what you can test and what happens in the real, messy world. He accepted that perhaps testing could be blamed for failing to point out to management that serious problems can occur in the things that are not tested.

Testing is an act of prediction, and some things are rather less predictable than others. Deterministic systems ultimately must interact with highly unpredictable people. And when a system relies on its success by interacting with many people, all of whom are operating in a new environment and faced with the melee that is the travelling public, it is unpredictability that is beyond the best of crystal balls.

Newsletters

Sign up to our Weekly Round-Up for the past week’s not-to-be-missed news analysis and insight delivered straight to your inbox.