New performance standards for voyage data recorders, VDRs, come into force today. While the measures, which apply to newbuild and retrofits,s under the IMO’s MSC.333(90) may resolve many of the concerns expressed by maritime accident investigation agencies over the years here remains a way to go before the situation is satisfactory.
In 2008 the UK Marine Accident Investigation Branch, MAIB reported in a preliminary investigation of and incident involving the ferry European Endeavour: “Once the situation was stabilised and the vessel had been made safe, the VDR save function was operated. However despite the unit being fully functional, the accident data was not saved. This occurred because the operators were not familiar with this type of VDR. Also, disk capacity had been poorly managed by VDR service representatives, resulting in reduced data storage capability.”
Following another incident Germany’s BSU commented:Says the BSU report
In 2010 Paul Drouin of Safeship, a private sector maritime investigator says in a comment on the LinkIn Maritime Accident Investigation Group: “…the issue has come up in quite a few accident reports world-wide. In Canada there was the Cast Prosperity/Hyde Park collision (TSB report M05L0205) where bridge audio on the Cast Prosperity was very poor. This echoes German findings such as those found in BSU Report 343/04.
“In the Canadian report we can read that ”… although the VDR had been tested upon installation and found operational, the quality of the recording was not verified under actual or realistic working conditions, nor was this required by the standard. This precluded confirmation that the in-service recordings would be of sufficient quality to effectively aid marine investigations.”
Under the new standards, among other changes, recording times will be increased to thirty days. A common problem has been seafarers who haven’t save data following an incident. The thirty day recordings should largely make action to save data unnecessary. Audio recording will be improved. Specialist technicians will not be required in order to test that a device is working.
Data must be stored in a fixed protective capsule capable of storing data for up to two years. There must also be float-free ‘medium’ capable of storing data for six moths and equipped with a locator beacon. Finally, there must be a long-term recording available easily from within the vessel.
Although these measures will certainly enhance the chances of investigators getting their hands in usable data they don’t address to issue of software standards that can make it difficult to translate that data into usable form in a timely manner.
VDRs are in much the same state as home computers in the 1980s with many brands requiring proprietary software to make them run. In the end customers voted with their wallets to bring, for instance, the number of operating systems down to the three main versions used today. This is unlikely tohappen with VDRs.
In a paper on digital forensics published in 2o13 Mario Piccinelli and Paolo Gubian of the University of Brescia, who recovered data from the Cost Cncodia’s faulty VDR,said: “One of the most significant challenges in digital forensic investigations on VDRs is that the data is not supposed to be readily available to the investigators in the native format they have been recorded in on the medium. Standard protocols provide that the FRM is opened by a technician sent by the manufacturer, who then extracts the data with proprietary software and gives the investigators processed data in form of a video replay of the accident, sound files or text files with timestamps and values from the sensors. As of now, the procedure itself cannot be validated by third parties because of the intermediate elaboration of data by proprietary software. The current IMO Performance Standards for VDR/S-VDR do not specify a recording format for VDR data. Consequently VDR manufacturers have adopted various approaches and methods. While widely accepted, this procedure is far from being forensically sound.
“In the (Cost Concordia) case … the VDR manufacturer provided the authors with a generic version of the replay software, which did not show some values that were needed for the investigation such as the status of the watertight doors, making it necessary to work around this issue by writing customized replay software. At the same time, the investigators appointed by the court received full support by the manufacturer and were thus provided with the more complete software. This inconsistency raises questions of fairness when defendants are completely reliant upon the VDR manufacturer in order to retrieve data for analysis.”
With the advances made in MSC,333(90) perhaps standardised software and data format should be next on the VDR agenda.