Heathrow Terminal 5
Went to a great seminar in Civil Engineering on T5: History in the Making. The speaker was Andrew Wolstenholme, BAA's Project Director for Terminal 5. (His uncle was "voice of the 1966 World Cup final", Ken Wolstenholme!)
The importance of T5 to BAA's future is simple to articulate: without T5, there's no Heathrow, and without Heathrow, there's no BAA.
Andrew had great pictures of Heathrow from 25,000 feet - "Terminal 5" is actually the main building on the hard west side of the airport, while Terminals 5A and 5B are standalone buildings (Atlanta-style) on the apron just to the east. The whole T5 project is about as big as the whole T1+2+3 complex put together, and will cost £4.2bn in total.
The main issues I noted were change, risk and safety. Andrew spent his first six months just studying other big projects to find out why costs overran, why delivery was late, or why there were fatalities. The main reasons for late delivery were that people didn't understand the risks and they underestimated the impact of change. "Change should be considered as a norm", he said. Software people have known this for some time!
It turns out that the "usual approaching to contracting" on construction projects is for the client (e.g. BAA) to write a contract that transfers all the risk to contractors, with a price determined in advance. This means that contractors' profits are whatever they can get, so they'll "trade" on gaps in the design spec, i.e. do the cheapest possible thing. BAA's "T5 Agreement" puts all the risk on BAA, as they reimburse "proper" costs and promise a pre-arranged profit to contractors.
This raised a fair few questions - "When did BAA become a registered charity?" - but Andrew's position was essentially that anything that gets T5 built on time is worth it. The idea is that a contractor's project manager can go to his board of directors and show a guaranteed profit, so the board will be disposed towards working hard for BAA.
BAA is unusual in being able to afford to take on such big projects and manage its contractors the way it wants. I detected a certain corporate imperiousness that you often get with the successful privatised industries. (BA and BT, this means you too.) It's essentially: Yes, we're public companies now, and consumers don't have to choose us, but let's face it, they will.
So that's risk covered, and it feeds right into the next big idea: "integrated teams". Get as many people from the client and contractors and sub-contractors together as early as possible to walk through the spec. The spec isn't locked down at that time. I was just thinking about Susan Eisenbach's "Scope - Quality - Cost - Time" factors for software engineering (somehow, Cost has become Resources in the current slides!) when Andrew put up a slide about "Making T5: Safe - Quality - Within budget - Bang on time" !
Andrew discussed some examples of how BAA is "getting it right" on T5: logistics (one vehicle comes through the gates every 40 seconds for four years), the roof lift for the main building, installing the new control tower, and the mechanical fit out for the main building. Over and over, the lesson was that it's incredibly valuable to practise off-site first. In the case of the roof lift (for the largest single-span roof in Europe), their trial in Yorkshire yielded 126 recommendations that would otherwise have been learned on-site and affected the whole project's "critical path". In the case of the fit-out, large concrete structures were pre-fabricated in Scotland "in a safe factory environment" (i.e. safer than a construction site) and shipped down. The aim was to turn the T5 site into an "assembly line", with components arriving "just in time", and it looks like they largely succeeded.
And when the 12m-high steel rings that comprise the control tower didn't quite fit together properly, finding the right technical solution was the only thing that mattered. BAA wrote off the cost of the bad rings, in line with the risk-taking ethos of the T5 agreement I mentioned earlier. The software analogue would be to discover that RDBMS 1 isn't scaling like it was supposed to, so the client consciously pays to replace it with RDBMS 2, whatever the cost. Doesn't Usually Happen, does it?
Something else that really hit me as a massive difference between construction and software projects is the safety factor. Software for even life-critical tasks doesn't kill the programmer. On T5, they were aiming for zero fatalities, but had one last year. However, industry metrics would suggest they "should" have had five by now.
Another vignette: after 21 months and £1bn spent, BAA's CEO visited the site - and couldn't see anything. Andrew spent that much just on the underground transport infrastructure - the Heathrow Express extension, the Piccadilly Line extension, service road tunnels, the famous diversion of two rivers. These aren't "behind the scenes" tasks. They're "plumbing" that create a platform for the rest of T5. Software projects also rely on important but largely invisible work to define a platform for the user application. I'm thinking of things like specifying, building and testing servers; choosing application and messaging frameworks; coding auxiliary programs like test transaction generators and data converters. But software projects rarely spend 25% of their budget on platform-building.
Andrew believes that mega-clients like BAA have a responsibility to "set the environment" in the UK construction industry. If BAA demand that their contractors practise key tasks off-site, then those contractors will develop skills that benefit smaller clients who can't afford to dictate terms. I'm intrigued by the definition of a small client as "someone who wants a £20m office park built"; a small software project is about three orders of magnitude smaller!
Heathrow Terminal 5 opens 4am, 30th March 2008.
The importance of T5 to BAA's future is simple to articulate: without T5, there's no Heathrow, and without Heathrow, there's no BAA.
Andrew had great pictures of Heathrow from 25,000 feet - "Terminal 5" is actually the main building on the hard west side of the airport, while Terminals 5A and 5B are standalone buildings (Atlanta-style) on the apron just to the east. The whole T5 project is about as big as the whole T1+2+3 complex put together, and will cost £4.2bn in total.
The main issues I noted were change, risk and safety. Andrew spent his first six months just studying other big projects to find out why costs overran, why delivery was late, or why there were fatalities. The main reasons for late delivery were that people didn't understand the risks and they underestimated the impact of change. "Change should be considered as a norm", he said. Software people have known this for some time!
It turns out that the "usual approaching to contracting" on construction projects is for the client (e.g. BAA) to write a contract that transfers all the risk to contractors, with a price determined in advance. This means that contractors' profits are whatever they can get, so they'll "trade" on gaps in the design spec, i.e. do the cheapest possible thing. BAA's "T5 Agreement" puts all the risk on BAA, as they reimburse "proper" costs and promise a pre-arranged profit to contractors.
This raised a fair few questions - "When did BAA become a registered charity?" - but Andrew's position was essentially that anything that gets T5 built on time is worth it. The idea is that a contractor's project manager can go to his board of directors and show a guaranteed profit, so the board will be disposed towards working hard for BAA.
BAA is unusual in being able to afford to take on such big projects and manage its contractors the way it wants. I detected a certain corporate imperiousness that you often get with the successful privatised industries. (BA and BT, this means you too.) It's essentially: Yes, we're public companies now, and consumers don't have to choose us, but let's face it, they will.
So that's risk covered, and it feeds right into the next big idea: "integrated teams". Get as many people from the client and contractors and sub-contractors together as early as possible to walk through the spec. The spec isn't locked down at that time. I was just thinking about Susan Eisenbach's "Scope - Quality - Cost - Time" factors for software engineering (somehow, Cost has become Resources in the current slides!) when Andrew put up a slide about "Making T5: Safe - Quality - Within budget - Bang on time" !
Andrew discussed some examples of how BAA is "getting it right" on T5: logistics (one vehicle comes through the gates every 40 seconds for four years), the roof lift for the main building, installing the new control tower, and the mechanical fit out for the main building. Over and over, the lesson was that it's incredibly valuable to practise off-site first. In the case of the roof lift (for the largest single-span roof in Europe), their trial in Yorkshire yielded 126 recommendations that would otherwise have been learned on-site and affected the whole project's "critical path". In the case of the fit-out, large concrete structures were pre-fabricated in Scotland "in a safe factory environment" (i.e. safer than a construction site) and shipped down. The aim was to turn the T5 site into an "assembly line", with components arriving "just in time", and it looks like they largely succeeded.
And when the 12m-high steel rings that comprise the control tower didn't quite fit together properly, finding the right technical solution was the only thing that mattered. BAA wrote off the cost of the bad rings, in line with the risk-taking ethos of the T5 agreement I mentioned earlier. The software analogue would be to discover that RDBMS 1 isn't scaling like it was supposed to, so the client consciously pays to replace it with RDBMS 2, whatever the cost. Doesn't Usually Happen, does it?
Something else that really hit me as a massive difference between construction and software projects is the safety factor. Software for even life-critical tasks doesn't kill the programmer. On T5, they were aiming for zero fatalities, but had one last year. However, industry metrics would suggest they "should" have had five by now.
Another vignette: after 21 months and £1bn spent, BAA's CEO visited the site - and couldn't see anything. Andrew spent that much just on the underground transport infrastructure - the Heathrow Express extension, the Piccadilly Line extension, service road tunnels, the famous diversion of two rivers. These aren't "behind the scenes" tasks. They're "plumbing" that create a platform for the rest of T5. Software projects also rely on important but largely invisible work to define a platform for the user application. I'm thinking of things like specifying, building and testing servers; choosing application and messaging frameworks; coding auxiliary programs like test transaction generators and data converters. But software projects rarely spend 25% of their budget on platform-building.
Andrew believes that mega-clients like BAA have a responsibility to "set the environment" in the UK construction industry. If BAA demand that their contractors practise key tasks off-site, then those contractors will develop skills that benefit smaller clients who can't afford to dictate terms. I'm intrigued by the definition of a small client as "someone who wants a £20m office park built"; a small software project is about three orders of magnitude smaller!
Heathrow Terminal 5 opens 4am, 30th March 2008.
posted by Alex Buckley at
11:01 pm
0 Comments:
Post a Comment
<< Home