The government’s recent decision to split the production of Offshore Patrol Vessels (OPVs) has been met with a mix of delight and dismay. Delight by those with an interest in the shipyards in Adelaide SA and Henderson WA, dismay by those who missed out. There was also more than the usual grumbling by those weary of multi-billion dollar defence projects being used to buy votes. But while the timing of the announcement was politically significant, in terms of the advice the government has been getting, it makes perfects sense to split production. Here’s the story.
In late 2014, the government commissioned the RAND Corporation to report on Australian naval shipbuilding at a cost of $2.5 million. Released last year, the report examined the challenge of building frigates to replace the Anzac class when they leave service from the mid-2020s onwards. Using a black box model designed to capture the process of hiring and training shipyard personnel, RAND estimated the cost and schedule impact on the frigate program due to the looming gap between the conclusion of the Air Warfare Destroyer (AWD) program and the commencement of the frigate program. Put simply, they estimated the cost and schedule impact of the infamous ‘valley of death’.
RAND then analysed options to fill the gap, including building three, four, or five OPVs commencing in 2017. Their insight was that the labour cost of building the OPV could be offset by the reduced cost of the frigate program due to the reduced number of ‘unproductive’ labour hours. As they put it, taking a four OPV build as an example:
‘In essence, the four OPVs could be built basically for “free,” given that they are sustaining productive labor that reduces the costs of unproductive labor when building the workforce for the Future Frigate construction.’
In addition, RAND predicted that building OPVs will yield substantial improvements to the schedule performance of the frigate program.
For reasons best known to the authors of the RAND report, the results are presented in an obtuse way that makes it difficult to unpack the details. But after a bit of reverse engineering (the details of which will be laid out in next month’s 2016 ASPI Budget Brief), it looks as though the RAND model forecasts savings in the frigate program of around $100‑250 million from building two to five OPVs in Adelaide. The exact figure depends on a number of factors, including the unit cost of labour and the number of OPVs.
The RAND report didn’t consider a split program, and the claimed savings will only survive if the additional cost of building OPVs in two locations is less than the savings in the frigate program. That looks to be the case. As best I can estimate, duplicating the OPV production line results in additional labour costs of around $20‑50 million, depending on unit labour costs and the numerical split of vessels between the two sites. Even doubling that figure to take account of duplicated non-labour start-up costs still leaves room for a saving. So on the basis of the RAND report, there’s a prima facie business case for building at least some of the OPVs in Adelaide. But that might be easier said than done. There are a number of countervailing factors the RAND modelling doesn’t take into account.
First, it’s unlikely that a single firm will operate yards in Adelaide and Henderson. So who’ll be responsible for building the initial OPV(s) in South Australia? To provide workforce continuity—the source of RAND’s purported savings—it will probably have to be the SA shipbuilder. If so, who’ll manage the contracts with suppliers and subcontractors? We could end up with two separate supply chains and two sets of management and engineering overheads for the OPV program.
Second, because different firms have been shortlisted for the OPV and frigate designs, there’s a risk that the Adelaide yard will have to adapt to three different design-production schemes in quick succession as they go from AWDs to OPVs to frigates. Don’t forget that it’s been necessary to bring in a shipbuilding management team from Navantia to get the AWD project back, in part at least, because of problems arising from a mismatch between the design and production methodologies.
Third, the RAND report assumed that the submarines would be built offshore. We now know that all twelve vessels will be built it Adelaide. So around the end of this decade we’ll have no less than five programs (potentially undertaken by five different firms) competing for space, labour and facilities at the Adelaide yards: Collins maintenance and upgrades, the future submarine, the future frigate, the conclusion of the AWD program and the initial OPV build. The competition for skilled space, facilities, labour and technical expertise may become fierce.
Notwithstanding all the potential problems with a split build, surely it’s worth a go if there’s a saving of up to $200 million to be made? That depends on how much faith you put on the RAND modelling. As I explained last year, the RAND modelling rests on shaky foundations (in fairness, here’s RAND’s response). The RAND model begins with a production schedule that’s doomed to fail, and then simulates its execution by managers with zero foresight. By its very construction, the RAND methodology will unavoidably yield a more pessimistic picture of schedule slippage and cost escalation than reality. Consequently, the true extent of potential savings from the frigate program is likely to be less than predicted by RAND. Or to put it another way, the OPV may be an unnecessary solution to an exaggerated problem.
The government is taking a big gamble by splitting the production of OPVs between Adelaide and Henderson in pursuit of the uncertain schedule and cost savings promised by RAND. The question of schedule slippage has been rendered largely irrelevant because continuous production delays the delivery of vessels to the point where a life-of-type extension for the Anzacs is necessary anyway (hence the sensor, weapons and combat system upgrades for the Anzacs in the 2016 Integrated Investment Plan). And even if the savings are as large as promised, they only amount to around $200 million out of a $38 billion program. Is it really worth complicating the execution of these massive programs for the sake of saving 0.5% of the total cost?