1. How does AI improve tank wash scheduling and optimization?
AI-powered decision algorithms can help reduce the cost a bulk fleet incurs resulting from tank washes. Advanced statistical methods typically included under the umbrella term Artificial Intelligence can improve the outcomes an optimization algorithm generates to reduce the cost of tank washes. These methods include machine learning, neural networks, or even large language models. They may be used to estimate certain parameters or adjust constraints in an optimization algorithm also known as a solver.
Today most people use tribal knowledge to optimize decisions (such as schedules, routes, assignments) for tank trailer washing. Tribal knowledge, albeit invaluable, comes with the challenge of being biased, error-prone, and not portable as staff retire or leave a company.
Bulk fleets with tank truck trailers are unique because trailers transport goods without packaging. Therefore, a commodity can be contaminated by another commodity that was loaded on a trailer before. Many shippers therefore require tracking the substances previously contained on a trailer for multiple substances. This is commonly referred to as tracking last-contained or priors. If substances loaded on a trailer sequentially are not compatible, a trailer will need to be washed out.
Washes are expensive because they involve the cost of the washing, the cost of moving the trailer to and from a wash location, as well as the opportunity cost of the trailer and driver not being available. Practically, the cost of washes can be reduced first by smartly assigning trailers to loads, reducing the number of total washes required, and secondly by reducing the cost associated with a wash at a certain location and the cost of traveling to and from such location. As prior and following loads need to be considered, the problem involves sequential decision-making.
Optimization solvers for sequential decision-making require estimating certain parameters, say an Expected Time of Arrival (ETA) or Projected Time of Availability (PTA) or the demand or rate for transportation for a certain origin-destination (OD) pair. Finally, flexibly defining, and relaxing constraints are usually a painful process for optimization solutions that require much custom engineering. For all of these, the opportunities new or better-performing AI techniques provide are welcome solutions.
2. What is the best system for managing last contained materials?
The best system to manage last contained materials is a specialized, modern transportation management system (TMS) that possesses real-time management capabilities and is built for A.I. Specifically, the TMS needs to be able to host compatibility matrices, automatically track last contained commodity and compatibility when a dispatcher assigns a trailer that results in top-loading one product on the other. Most importantly, such a system needs to have advanced alert capabilities to advise dispatchers and management of wrong decisions. The TMS needs to seamlessly integrate into real-time trailer tracking and have native yard management capabilities. The TMS should have deep bulk transportation capabilities, ideally in one of the subspecialties
(food bulk, chemical bulk, construction bulk, etc.) that is relevant for the fleet. Finally, if the fleet intends to use AI to optimize trailer washes, the TMS needs to have a real-time integration layer that aggregates all necessary data and AI capabilities to improve tank wash optimization outcomes.
Most tank truck fleets today manage last contained materials in a system composed of spreadsheets and use the dispatcher's tribal knowledge to avoid contamination or manage washes. Although this traditional system is almost free, it is prone to error and biases. While the cost of biased decisions is usually just paid by the owner in the form of low efficiency, the cost of an error for tank truck fleets can be substantial: either as it contaminates and makes unusable another substance or as it – even worse – creates an accident if two chemicals are reactive.
Older TMS systems usually struggle with managing last contained materials because they have poor or only expensive integrations into adjacent technologies (e.g., trailer tracking), have poor dispatch planning interfaces with advanced alert capabilities, and do not offer natively embedded optimization algorithms that can help reduce the cost of inefficient last-contained management.
Therefore, the best option is a modern transportation management system (TMS) that specializes in bulk, possesses real-time management capabilities, and is built for A.I.
3. How can private fleets use AI to compete with for-hire carriers?
Private fleets compete with for-hire carriers in two areas.
Firstly, private fleets need to defend themselves against the possibility of being replaced by for-hire fleets, i.e., the private fleet needs to justify its existence against the alternative of outsourcing transportation to either a resolute for-hire carrier, contract freight, and spot-brokered model. Private fleet owners typically look to the fleet's ability to offer better customer service and cost efficiency. Artificial intelligence technology can help improve both customer service and cost efficiency as it gets embedded in optimization algorithms that improve dispatch decision-making. For instance, in an optimization solver, large language models (LLMs) can be used for dispatchers to flexibly modulate constraints and improve how viable its recommendations are in real life, enhancing overall adherence and effectiveness. AI can also be embedded in models to estimate certain parameters (say a Projected Time of Availability for a truck) involved in deciding whether a load should be tendered to the private fleet vs. an outside carrier.
Secondly, private fleets can compete in the for-hire fleet market, taking on loads from other shippers or brokers. They can leverage AI to identify those markets and shippers in which they can quote rates that are both competitive and improve the private fleet's utilization, effectively reducing the cost of having a private fleet to the shipper. AI techniques are usually embedded in freight network analyses and simulations to estimate the demand for, supply of, or rates associated with certain lanes (O-D pairs).
4. What is the advantage of using AI for bulk load optimization?
Bulk load optimization is different from traditional dry-van or refer optimization because it involves managing last-contained (priors) and tank washes. Most companies use tribal knowledge, spreadsheets, and, at best, a legacy TMS.
However, the most effective method involves AI techniques that are natively embedded in the TMS. The advantage of this lies in:
· Automation: Reducing costly errors that spreadsheets and humans are prone to. AI can be helpful in predicting root causes of errors (e.g., delays) that can then help alert users of errors.
· Optimization: Estimating parameters with better accuracy. Certain inputs in optimization can benefit from AI techniques to estimate, say, the demand for freight on a specific lane or its rate potential across the year. Finally, AI can be used to help flexibly modulate constraints that are incorporated into an optimization algorithm.
AI techniques are only as good as the data we provide them. Therefore, the most powerful AI systems live in a transportation management system that has a deep data integration layer and offers user experience where improvement recommendations can be presented natively in the TMS interface without switching.
5. How do I ensure compliance and visibility in hazmat or bulk transport?
Compliance and visibility in hazmatv or bulk transportation tie directly to driver management and driver activities. Traditionally, these activities get managed with a mixture of training, standard operating procedures (SOPs), and paper forms. Visibility is at best available after the fact.
Today, digital driver workflow management via the ELD offers the most significant opportunity to improve hazmat safety compliance at the driver level. While the ELD or a cell phone usually offers the driver a digital interface, the orchestration of such activities can be automated in the transportation management system. Modern transportation management systems (TMS), like BeyondTrucks, offer advanced driver workflow management capabilities, hosting the logic on when to trigger what workflows. The seamless interface with the mobile device in the cab allows for enforcing adherence to these activities in the cab, whilst collecting data in real time from the cab and sending it back to the TMS. TMS driver workflow capabilities therefore improve both hazmat compliance and visibility.
Hans Galland is the founder and CEO of BeyondTrucks, a vertical SaaS provider of Transportation Management System (TMS) solutions. A serial entrepreneur in real estate, investment banking, and logistics, he received the Gold Stevie Award for Best Entrepreneur in Transportation for advancing AI-native innovation in trucking.