Predictive analytics is commonly defined as the use of data, statistical algorithms and machine learning techniques to identify the likelihood of future outcomes.
Using new technologies to enable enhanced collection, storage and analysis of historical data, shipping companies are able to make predictions to enhance their operations. For example, sensor data can be used to identify which areas require priority in terms of maintenance, reducing the amount of time they need to be dry-docked and preventing delays.
Nevertheless, predictive analytics is also being increasingly used to speed up various steps in the supply chain. From a shipping perspective, certain ports are beginning to invest in predictive visibility projects as a means of keeping cargo moving when ships come in to dock.
“If one can more accurately predict when a ship arrives (or even better, when a specific container gets unloaded from a ship), many of the processes needed to move this container from the port to the final customer can be planned ahead of time,” says Dr. ManWo Ng, assistant professor of maritime and supply chain management at Old Dominion University in Virginia, US.
“For instance, terminal resource needs can be optimised, motor carriers can be better informed and containers can be picked up sooner, clearing congestion in container yards.”
The growing need for data analysis
For maritime, the concept of predicting the future is not a new one. According to Ng, terminal operators “have done forecasting for as long as there have been seaports”. Nevertheless, certain technologies, such as radio-frequency identification (RFID) and differential global positioning systems (DGPS), have facilitated the collection of more accurate real-time information. This enables a more precise prediction of when cargo will be available for pickup.
“Of equal importance are the developments in hardware and predictive analytics technology, from data warehousing to parallel computing, new predictive models, algorithmic advances and data visualisation,” says Ng. “All these have contributed to making predictive analytics more practical and accessible.”
A combination of factors has led to a need to boost cargo efficiency at ports. Favourable economic conditions have led to an increase in demand for cargo shipping. A side effect of this is that ocean carriers have steadily gotten bigger to accommodate more cargo. But Ng claims that these so-called megaships are putting unprecedented pressure on ports’ cargo handling capability.
“Another major development is the forming of alliances among container carriers, which has significantly complicated the inter-terminal logistics between the different terminals in a port complex,” he says. “All these (and other) realities have not made cargo handling easier at today’s seaports.”
The Port of Long Beach: a case study
Situated in Southern California, the adjoining ports of Los Angeles and Long Beach are the busiest in the US. With staggering amounts of cargo to handle, congestion can prove an issue, with lines of vessels at anchor waiting for space to berth.
“US ports, like many others around the world, are struggling to maintain or hopefully increase terminal productivity as ship call sizes continue to rise,” says Allen Thomas, chief strategy officer at Advent Intermodal Solutions. “We simply can’t do that without leveraging technology to help.”
The International Transportation Service (ITS) and Advent recently launched a predictive visibility solution at ITS’s port terminal at Long Beach. Over the past two years, the companies have developed a data exchange system that gives shippers, beneficial cargo owners (BCOs) and motor carriers the ability to see when cargo will be discharged five days before a vessel arrives.
The crux of this is an expansion to Advent’s eModal solution, an online portal currently being used by terminal customers to manage truck appointments. The ITS terminal operating system can supply eModal with a projection of when a container will be available for pick-up. With this date and time in place, the solution can automatically make an appointment for truckers based on their preferences.
“This primarily helps the trucking company (specifically the dispatcher/scheduler) by not requiring them [to] continually check back in eModal to confirm container availability and then make an appointment,” says Thomas. “This entire step is eliminated, saving about 20 minutes per container on average while providing confirmed availability times five days in advance (which aids both trucker and terminal labour planning).
“[The system] should enable more aligned scheduling of trucks over longer periods which reduces congestion/wait times typically found when everyone shows up at the same time ‘hoping’ to pick up containers as they become available throughout the day.”
In a recent World Cargo News article, ITS CEO Sean Lindsay said that in the first week of use for the eModal system, the company had taken 150 requests for appointments and that this number had jumped to over 600 in four weeks.
Meanwhile, the Port of Los Angeles recently introduced a new information sharing portal that digitises maritime shipping data, meaning it can be made available to BCOs and supply chain operators. Launched at a few terminals during a pilot programme, the portal is now being expanded to include all shipping lines and terminal operators by mid-2018.
In the future, projected container availability times could be spread across the portal to various components of the supply chain. This would allow terminals to identify equipment and labour needs, truckers to plan dispatch schedules, and BCOs to prepare their facilities for delivery.
Forecasting the future
Despite the benefits of predictive analytics, Ng highlights that potential issues remain. He claims that “objective and data-driven predictive analytics is the best in delivering consistently more accurate predictions in repetitive environments”. However, when operations change, historical data might not be of any use as these patterns will no longer be representative.
Even when a good predictive model is finally in place, it still requires periodic maintenance to learn from new scenarios. The Port of Virginia recently established a new department of innovation that employs analytics to support decision-making, according to Ng, but less forward-thinking ports will likely lack the human resources to cope with this new demand.
“Being from academia, I have to say that it is not easy to find graduates with both a solid understanding of the maritime industry as well as a technical background in predictive analytics.” says Ng.
In addition, each port will need its own tailored predictive analytics models, as no two ports are the same. This removes the potential for a company to create a common software solution that can be used at ports worldwide.
“While a predictive model is successful at port A, nothing guarantees that the same model will give useful predictions when applied at port B,” says Ng. “The dynamics at the ports are completely different.”
The Port of Rotterdam recently embarked on an initiative to develop a centralised dashboard application that will collect and process real-time water, weather and communications data. This will help the port make decisions that reduce wait times and determine optimal times for ships to dock, load and unload their wares.
For Europe’s largest port, this sort of investment makes a lot of sense. But in the conservative maritime industry, Ng contends, smaller ports are unlikely to adopt predictive technologies anytime soon.
“To compete for cargo, larger ports are under a lot more pressure to improve cargo handling velocity,” he says. “So, they are more likely to be forced to adopt it by the competitive business environment. But only time will tell.”