Grasping Fragility: ConexBird's first results

ConexBird is proud to announce that we have successfully produced preliminary results at a commercial location, as part of one of our ongoing pilot projects with major industry partners. The production of these preliminary results demonstrates both the viability of ConexBird’s system, as well as the necessity and urgency of having such a system in place.

During this trial, ConexBird’s algorithm found that 73% of containers did not match the structural ages indicated by their dates of manufacture. To understand this figure, it is important to note that:

‍

• Objects naturally lose their strength and durability through use due to material fatigue.
• Container shipping lines (and other intermodal operators) often rely on a container’s age as a proxy for its overall durability.
• This durability is a property of the container that exists independently of any specific damages it may have at a given moment.
• As a rule of thumb, containers under five years old can be thought of as “new” and “rugged”, while containers over ten years old can be thought of as “old” and “fragile”.
• Containers are subjected to variegated and heterogeneous conditions over their lifetimes: it stands to reason that they do not all degrade at the same rate.

‍

For this trial, ConexBird took vibration measurements of containers being handled normally at the test site over a specified period. These measurements provide a consistent, impartial means of comparing a given box’s structural condition to other containers: however, merely indicating that a container looks “better” or “worse” is not particularly useful without a reference point to attach the qualifier to.

This reference point is provided by a container’s “nominal age”: its date of manufacture. ConexBird’s machine-learning algorithm compared container vibration measurements both to each other and to their respective container ages, and thus determined approximately how old the container’s structure appears to be. This result includes the algorithm’s level of confidence in the assessment (expressed as a percentage). From the set of results we can observe the following:

‍

• When compared to their nominal age, 73% of containers were found to be either worse (“older than expected”)  or better (“newer than expected”) to a noticeable degree.
• Thus, relatively few containers have a structural strength which actually corresponds to their nominal age.
• Overall, containers are more likely to be in worse condition (“older”) than their nominal age would suggest.
• Our data indicates that this discrepancy is not confined to a single operator or group of operators: rather, it is an industry-wide problem.

‍

We would like to highlight two specific containers observed over the course of the trial, the details of which are summarized in the following infographic:

The first of these containers is a relatively new unit, manufactured within the past five years and thus ostensibly in good condition; ConexBird’s algorithm is 100% confident that this particular container is significantly weaker than its age would indicate, and that it resembles containers which are typically five years older.

The second example is that of a relatively old unit, manufactured in 2005; inversely, ConexBird’s algorithm is 100% confident that this container is significantly stronger than its age would indicate, and that it resembles containers which are typically ten years younger.

While promising, these initial results are merely a glimpse of what our system is capable of doing: ConexBird looks forward to sharing more details on our progress in the weeks and months to come. Together, we intend to change the future of container shipping, and we look forward to continuing this journey with you.