In the last post we arrived at a strategy which we like, this being:
- Start our operations sometime in March.
- Use two vessels, unfortunately we can get the two we’d like though, so we’re using a slight cheaper, slightly less capable vessel as our second.
- Modify the capable vessel so it can carry more foundation units per trip (3).
However, we saw that the division of labour we’d set wasn’t optimal and that we off hired one of our vessels early and lost some of the advantage gained by this strategy.
The simulation set up for this analysis is very simple. We’re using the simulation we developed last time and adjusting the number of foundations to be installed by each vessel, until we get to the point that both vessels are on hire throughout this installation. To start with change the 54/18 split to 48/24.
The work assignment for this scenario is shown below. Each row represents the simulation of one March start date and the colour blocks indicate the vessel that installed the foundation. This is an improvement as the less capable vessel off hires later in the project, but it’s still going early.
Let’s try modifying the simulation to have a 30/42 split. This process is very simple and is a good example of using Mermaid iteratively to optimise work.
The work breakdown shown above for a 30/42 split is good, the second vessel is on hire for most of the operation and our rate of progress should therefore be fairly high throughout. The results are presented below.
As suspected, the duration of installation is the shortest seen throughout this process, clocking in at a March median of 122 days. The rate of progress is fairly high and fairly linear and this is starting to look like an operation we might be happy to perform.
The cost is slightly raised by this division of labour from the previous option. However, we’re still around the lowest values we’ve seen, with the March median being £33.5 million.
The Next Steps
We’ve performed a fairly extensive analysis process during this series of posts and this represents the end point. We’ve defined an operation, set various constraints throughout the process and manage to find an optimum strategy for our installation work.
At this point its worth taking a quick look at the duration box-whisker plots generated in Mermaid for each of the vessels used in the 30/42 split scenario. These are presented below and we can see the type of seasonality we’d expect (the less capable vessel is presented first). We can see that April may represent a better start time than March (which we selected earlier), although the spread of results (high of box and whiskers), and therefore the associated risk, is less in March. If we were performing this analysis for a real operation we may we now dig into these start dates in more detail. We may also look to identify any potential bottlenecks and optimisations that we can remove or incorporate to improve our performance further. In short, we’d perform the operation multiple times in the safety of Mermaid until we were happy we had something we could reliably perform at sea.
Anyway, that’s probably a whole other series of posts.