When the K-141 Kursk submarine went down in 110m of water in the Barents Sea a year ago, the world looked on in horror. For days it was unclear what had caused the accident and the fate of the 118 men on board was equally uncertain. After nine agonising days, hope for the survivors was extinguished when Norwegian divers confirmed that the hull was full of water.
From then on, the rescue became a mission to salvage the submarine to minimise any environmental dangers from the broken vessel and to discover the true cause of its sinking.
Raising the Kursk was always going to be a horrendous task, but behind the scenes IT was to play a key role in the salvaging efforts.
The inhospitable waters of the Barents Sea made a salvage operation a nightmare, while the fragile state of the vessel worried engineers. Financially, the Russian government was struggling to find the funds for a rescue.
Diplomatically, the situation was delicate, involving a complex web of international relations and the touchiness of a Russian government that many of its people felt had failed to do enough to save the men or to communicate the true facts of the case.
In all, perhaps the most straightforward aspect of the whole affair was the technology. And yet the centrepiece of that was some beta software which most of the principal operators had scarcely even seen before.
A Norwegian company Halliburton Subsea was chosen by the Russian government to conduct a preliminary study of the wreck and how to salvage it, along with Dutch companies Smith International and Heerema.
Halliburton needed to create a comprehensive plan on how to bring the submarine to the surface in such a way that any environmentally hazardous material, such as nuclear fuel from the submarine's reactors, was rendered safe. And as far as possible, damage to the vessel was to be minimised, so that rescuers could form an accurate picture of what had happened to cause the sinking.
The Halliburton team decided the best way to do this was first to create a computer model. After considering several approaches, they opted to go with software from Autodesk, and asked the Autodesk dealer AGS in Norway to assist.
Birger Haraldseid, a consultant at Halliburton who worked on the project, explained what the company had in mind. "We wanted visualisation - we had to visualise the salvage operation as a concept initially, and then we had to use visualisation to further refine the process," he said.
Although the company works with undersea operations all the time, this one was rather unusual. Most of Halliburton's projects involve laying pipelines or cables - this was a badly damaged submarine. "It was a totally different animal altogether," said Haraldseid.
The company had to spend more time than usual working out the exact engineering requirements and visualising them carefully to ensure they were getting it right.
The team of software engineers were using Autodesk's Inventor 4.0 software, in a beta version. Although it was a beta, the software was chosen because it was cheaper than alternatives - at £6,000 for a seat (that is, £4,500 for the software and the rest on a PC), it came in at a third of the price of standard three-dimensional visualisation software, which typically costs between £15,000 and £20,000 (including the price of a workstation).
Many of the team involved had not worked with the software before, so some had to undergo intensive training. This worked out to be relatively straightforward, as the inexperienced engineers were guided by a group of experts from AGS who were familiar with the systems.
"It was quite easy to use in the end," reported Haraldseid. As the team was working closely with Autodesk too, beta glitches in the software could be swiftly sorted out or worked around.
First, the team got together a set of two-dimensional manual drawings of the submarine from the Russian navy, and set about digitising them and feeding them into the Autodesk software - first into Inventor and then into a related program called 3D Studio Max, a visualisation program sometimes used in the film industry.
This arduous process took 10 days but Halliburton reckoned it would have taken up to three months using alternative systems.
When this process had been completed, the team was furnished with the correct digitised construction blueprints of the finished submarine from the Russian navy. It turned out there were some significant differences between these blueprints and the original drawings. So these had to be incorporated into the computer model.
Fortunately, this turned out to be a simple matter of entering the new parameters and letting the system calculate the changes. Then the completed model could be used by engineers to work out how best to lift the Kursk.
"We did not model every single bolt and tap in the hull, as you can imagine, because that would have been too much, but we did go into very fine detail. We also modelled the crane and barge that were to assist in the lifting," said Birgit Pettersen, sales manager at AGS.
The Kursk, with 118 crew, was a big vessel - at 154m, more than twice the length of a jumbo jet, and with a displacement of 18,300 tons. Lifting it up involved a close understanding of the stresses and strains that would be placed on each part of the vessel during any lifting process.
For this precarious stage, the complex modelling that the software team had produced was invaluable.
The 3D model came in handy for other purposes too. When news crews from around the world converged on the site of the submarine's sinking to cover the salvage attempts, they found the computerised drawings useful for illustrating to viewers just what was going on under the water. The model appeared on numerous television programmes.
Even though the software was only in beta, it was surprisingly stable, Halliburton found. "We considered it an excellent tool," said Haraldseid.
In fact, the biggest problem, according to Haraldseid, was working with the Russian government. "The Russians were slightly more difficult to work with than we were used to in the western world. They were politically driven. And although the Russian Northern Fleet was calling the shots, it never signed a contract with us and never agreed a timeframe. It was changing the goalposts all the time," he complained.
In the end, the plan from Halliburton was not chosen to raise the vessel. The Russian government seemed to decide that it was too expensive and has opted for an alternative plan from a Dutch company. The £55m salvage effort is carrying on now, with British divers working alongside Russians to break the submarine apart and raise it in pieces.
The operation to raise the Kursk has taught all involved a valuable lesson not just in technology, but in the complexities of managing a fraught international situation with many partners involved, and all of it taking place under the intense gaze of the world's media.
For most IT professionals, it is not often that unseen beta software turns out to be the least troublesome aspect of a project.
Fiona Harvey writes for the Financial Times
This was first published in August 2001