- The risk of fire originating from dangerous goods in containers is well known.
- The container booking process can help guard against mis-declared or undeclared containers getting on board.
There are several reasons why DG cargo may be misdeclared or undeclared. Unintentional errors at time of declaration might be one explanation, but there are concerns that a significant proportion of incorrect declarations are deliberately made by the shipper, says an article published in NEPIA.
Advantage of machine learning
There are many systems in use that can detect and therefore prevent unscrupulous shippers placing such bookings. Manual systems require office staff to constantly scan the booking environment. However, recent technological advancements have helped to automate this process, taking advantage of machine learning.
We spoke to Chris Barker and James Douglas from Exis Technologies to find out more about the Hazcheck Detect (HCD) system, used by carriers that include Maersk, ONE and Hapag Lloyd.
How it works
Exis host HCD using application programming interfaces (APIs), which allows two applications to talk to each other. The container carrier decides when to trigger the API, which then screens the booking data for certain pre-defined keywords and rules hosted by HCD in an industry library. Carriers have the option to maintain private libraries in HCD for additional screening options such as wildlife, chemical or precursors.
This part of the process is critical because it can identify cargoes which have been deliberately or accidentally mis-declared that may pose a risk. The search can be a simple keyword, combination of words or rule applied against the cargo description.
Most HCD users will screen the booking both at the start of the process and when finalising the booking, when shipping instructions and dangerous goods declarations are provided. Many users choose to screen every time a change to the booking is made.
Instant detection
The user receives screening results or hits within a matter of seconds of requesting the API. These hits allow carriers to focus attention on bookings which are systematically identified as suspicious, and to identify suspicious cargo when last minute changes are made to the booking, bill of lading or shipping instructions as they occur.
HCD customer demands differ, so some use their own systems to manage hits and follow up whereas others use an additional feature of the HCD user interface to manage their hits.
Actions and alerts
HCD customers have dedicated DG teams to manage the follow up process. The team will review the hits detected by HCD, filtered by categories, regions, and anything else of relevance detected during the screening process. A decision will be made; including but not limited to the following actions:
- Pause the booking process
- Request more information from the shipper, such as safety data sheets or more detailed cargo descriptions
- Require a container inspection note
- Assign to DG specialist
- Cancel booking or request cargo is re-booked with correct declaration
- Allow cargo process to continue if there are no concerns
Information on confirmed hits will be provided back to Hazcheck Detect for the purpose of machine learning including fine tuning, and/or adding new keywords and rules to the industry library.
Industry library: form and function
To prevent rejected bookings being accepted by another carrier, Exis encourage the use of the same rules / keywords across all customers. This forms an Industry Library, where there are currently more than 3,000 rules.
The initial users of the HCD system have vast amount of experience in booking DG cargo using their own systems. This helped to create a consolidated Industry Library with the most effective and logical search terms.
The search terms in the Industry Library are all cargo related and can be used by all carriers. There is no information that could allow the identification of shippers. However private libraries dedicated to each carrier may have shipper’s search terms incorporated.
Smaller carriers who enter a service agreement with Exis will have access to the Industry Library. However, they will be asked to provide access to their library of cargo keywords and rules.
The Industry Library search terms are constantly evolving, and this is an important part of the continual improvement process.
A working example
To prevent unscrupulous shippers cheating the system, HCD is adding functionality to prevent them from ‘shopping’ around.
For example:
If Carrier A questions a shipper as the result of a suspicious hit from HCD, and based on feedback from the shipper, subsequently cancels the booking or the shipper cancels the booking to avoid scrutiny, HCD will flag the booking as being recently cancelled.
If that same shipment is offered to Carrier B for the same vessel, or any other vessel, this will be flagged on HCD. This provides an indication that the shipper is trying to circumvent the system by finding a carrier willing to accept without question.
These unscrupulous shippers will continue to try and load the cargo on other ships, so all carriers need to be alert to this and manage the risk accordingly.
Scale of bookings and detections
At the time of writing, Exis screens 25 million items per month and this is expected to grow. It is estimated that 30% of the container line market are presently using HCD.
Most importantly, containers are currently being stopped from loading vessels because of hits generated by HCD. Feedback from one carrier advises 40 to 50 containers per week are stopped.
Did you subscribe to our daily Newsletter?
It’s Free! Click here to Subscribe
Source: NEPIA
