Quantum computing use case detailed Largest U.S. port doubles operational efficiency

While the Port of Los Angeles is the busiest port in the United States (number one in the nation in terms of throughput), it is also the quintessential inefficient port. In this era of global supply chain challenges, it is now more important than ever that ports perform at maximum capacity with maximum efficiency. With the surge in procurement caused by the epidemic, ports around the world are struggling to keep pace with rapid operations. Technology to optimize port infrastructure capacity is critical.

 

As a result, in January 2022, the Port of Los Angeles adopted quantum computing technology for the first time. Recently, data analytics firm SavantX has once again analyzed the operational efficiency of the Port of Los Angeles in the form of a report.

 

 

01Introduction: SavantX and Pier 300's HONE (Super Optimized Node Efficiency) Technology

 

 

The Port of Los Angeles and the neighboring Port of Long Beach handle more containers per call than any other port complex in the world: with more than 10.9 million TEUs (twenty-foot equivalent units) passing through each year, Pier 300 ranks in the top four of this gateway. Since July 2019, Fenix Marine Services, operator of Terminal 300, has been working with quantum computing and data optimization company SavantX to improve the efficiency of the terminal; focusing primarily on improving the efficiency of unloading containers from ship to shore, as well as the efficiency of deliveries from local import yards.The technology implemented by SavantX at Terminal 300 is known as the HONE (Hyper Optimization Nodal Efficiency) project, which uses proprietary data science and quantum computing to provide a real-time optimization solution for port usage.

 

What is HONE?HONE is an optimization engine. It is not simply a piece of software installed in the cloud or at the terminal, but a customizable Software as a Service (SaaS) that leverages the knowledge of Subject Matter Experts (SME), data scientists and quantum physicists to deliver cutting-edge solutions critical to optimizing global supply chain infrastructure.

 

HONE's goal is to improve efficiency by working with customers to identify key performance indicators (KPIs) and then optimize them. Optimizing KPIs can cover a wide range of possibilities ...... from the simple re-painting of truck parking area lines to the complex use of quantum computing to determine the best path a rubber tire gantry (RTG) should take to move a string of containers.

 

HONE excels at solving these problems because SavantX uses experts to work with customers to produce unique solutions that fit their use cases.

 

02Case Study: 3 Years of Results for Terminal 300 and HONE

 

 

In late 2019, SavantX began the initial implementation of HONE at the Port of Los Angeles Terminal 300.

 

SavantX's goal is to implement a process system that will increase import yard throughput while reducing truck wait times. Several key performance metrics have been identified. To measure throughput, daily unloads and deliveries were identified as key KPIs; other KPIs identified to improve terminal efficiency were: RTG movement distance, reservation ratio, productive versus unproductive movements, and average truck wait time.

 

With these KPIs in mind, the HONE solution tailored to Pier 300 was deployed in a phased approach over a period of more than 3 years, validating each new process while winning the approval of the operations department. Arguably, the KPI with the highest visibility is truck waiting time: HONE has significantly reduced truck waiting time and increased productivity while making Terminal 300 a more desirable terminal service.

 

03Maximizing throughput: balance between unloading, delivery

 

 

The purpose of the Pier 300 facility is to unload containers from the ship (discharge) and store them until the shipper sends a truck to remove them from the import yard, store them, and until the shipper sends a truck to remove them from the import yard (delivery). There are several limiting factors that can be controlled.

 

The first factor is unloading the ship. On average, a ship can hold over 4,000 containers. Since it takes 3 minutes to unload a container from a ship, then even with multiple ship-to-shore (STS) cranes, it takes several days of work to unload a ship.

 

The second limiting factor is the import delivery lead time. Deliveries depend on the coordination of 300 dock workers at the import yard and truck drivers who are not employed at the terminal. Truck drivers make appointments for shippers to pick up containers. Efficient delivery of containers depends on a solid strategy between yard workers and truck drivers.

 

A final limiting factor is the import yard itself. Containers may stay in the yard for hours, or weeks, before being delivered. Import yards have extremely limited space. An empty yard means that not enough containers are unloaded from the cargo ship and the yard is underutilized, while a full yard means that too few containers are picked up and therefore the yard becomes congested and difficult to operate.

 

Therefore, it is imperative to find a balance between unloaded and delivered containers: to maximize throughput, both unloading and delivery should be maximized while keeping the yard in balance.

 

04Data analysis

 

To determine how Pier 300's throughput has changed over time, the chart below records the daily number of discharges and deliveries for the period November 2019 to July 2022.

 

Figure 1 Daily unloading volume at Pier 300 with 15-day and 120-day moving average trend lines

 

Because the productivity of the yard varies considerably from day to day, two types of trends were identified. The first trend is a 14-day rolling average (Figure 1, solid line), while the second trend is a 120-day rolling average (Figure 1, dashed line). It should be noted that a partial data interruption occurred between February and April 2021.

 

This disruption did not affect the HONE operations and terminal efficiency. Figure 1 shows the increase in daily unloading over time throughout the implementation of HONE. This increase began in May 2020, shortly after the initial first phase of HONE began. By the fall of 2021, the trend in daily unloading volumes began to level off; currently, daily unloading volumes are similar to the fall 2021 figures. Over the 17-month period from May 2020 to October 2021, daily unloads have more than doubled on average.

 

Figure 2 Daily deliveries at Pier 300 with 15-day (solid line) and 120-day (dashed line) trend lines for the period October 2019 to July 2022

 

Daily deliveries (Figure 2) behave much like discharges. a positive trend begins to emerge in early 2020, eventually leveling off in the fall of 2021. It is easy to see the trend in these two figures, but to capture the change in deliveries and unloads, it is best to break it down by month.

 

Figure 3 Ridge chart of deliveries by month

 

Figure 3 shows a spine plot of monthly deliveries since the beginning of HONE. The graph takes the daily counts for each month and then creates a distribution from those values to help capture the distribution of counts for the entire month. The highest points on a month's distribution curve represent the most likely values for that month, while the values at the tail of the curve are unlikely to occur. This makes comparisons between months easy. Using Figure 3, it is clear that deliveries increased significantly after the implementation of HONE in early 2020. In the first year, HONE helped to double the average daily deliveries at Terminal 300, which allowed the terminal to efficiently handle the increased demand from the overloaded global supply chain during that period.

 

Long-term yard stability is another important factor. If the yard is not stable, unloading and delivery rates cannot be sustained. This means that neither can outpace the other over an extended period of time. If the unloading volume is much greater than the delivery volume, then the yard will eventually reach capacity and ships will not be able to unload until there are enough containers to deliver. If the opposite is the case, and delivery volumes dominate, then the yard will be emptied, which leads to inefficient use of the import yard. Both scenarios will inevitably lead to a potential loss of revenue for the terminal. For this reason, it is important to ensure that HONE keeps the yard stable as the number of daily unloads and deliveries increases.

 

Figure 4 Daily Variance between Unloading and Deliveries at Pier 300

 

Stability can be determined by simply looking at the variance between unloads and deliveries on a given day. This effectively shows the extent to which the yard is growing or shrinking (Figure 4). If the difference between the two is negative, there are more deliveries and the yard shrinks; if it is positive, there are more unloads and the yard grows. A stable yard will have a flat long-term trend: the 120-day trend (Figure 4, red dashed line) shows this.

 

 

Although unloading and deliveries increased over time, they grew at a similar rate, which kept the yard stable. This shows that while deliveries and unloading rates have more than doubled, HONE has helped keep the yard in a stable condition. If HONE's recommendations and optimizations remain intact, Terminal 300 should continue to see throughput figures 200% higher than before HONE.

 

05Optimization Strategy

 

The process of optimizing Pier 300 included optimizing the unloading and delivery of containers. Optimizing how vessels are unloaded faces many constraints as the vessels arriving at the terminal must be unloaded in a way that keeps the vessels balanced. To optimize unloading, HONE focuses on using historical data to help planners by providing unloading patterns based on previous visits of the vessel: these patterns provide planners with information on how containers should be unloaded to the terminal.

 

Optimizing delivery at the terminal involves two processes. The first is the development of a reservation system that controls when trucks can enter the terminal gate to pick up their containers, setting specific rules for when trucks can enter the gate.

 

1) All trucks must have an appointment to enter the gate unless an arrangement is made between the terminal and the trucking company.

 

2) Trucks may not arrive more than 45 minutes prior to the appointment time.

 

3) Trucks may not arrive more than 180 minutes after the appointment time.

 

These rules ensure that trucks enter the yard with the rubber tire gantry (RTG) in the vicinity of the container they are to pick up.

 

Enforcing these rules is essential to maximize yard throughput: by observing the difference between truck arrival times and appointment times, it is possible to determine whether the terminal and the trucks are complying with the appointment system.

 

Table 1 Basic descriptive statistics of appointment discrepancies

 

The data in the above table show that.

 

1) Trucks are on average 20 minutes late.

 

2) Half of the trucks arrive within the 70 minute window, starting 19 minutes before the appointment and ending 50 minutes after the appointment.

 

This means that the majority of trucks have complied with the appointment system since it was implemented: the smallest and largest arrival discrepancies do indicate that the terminal allows some trucks to violate the rules.

 

Figure 5 Average daily variance in appointment times and actual truck arrival times over a one-year period from July 2021 to July 2022

 

As with throughput, appointment variances can change over time. Figure 5 shows that appointment variances have remained stable for most of the past year, but a trend toward more late-arriving trucks has recently begun to emerge. Late arriving trucks are more detrimental to terminal efficiency because RTG must double back to pick up that truck's containers. This puts additional pressure on RTG, increases the waiting time for other trucks, and reduces the number of possible deliveries in a day.

 

The second optimization process for deliveries is to optimize the movement of the RTG. There are several reasons for this: The RTG is the backbone of the terminal delivery process. If RTGs have to be constantly repositioned, this is a time-consuming task because RTGs are not flexible machines and throughput will suffer. Like any machinery, the more it is used, the sooner it will need maintenance.

 

If an RTG breaks down and cannot be replaced quickly, the throughput of the terminal suffers greatly; therefore, minimizing movement while maintaining delivery means less downtime for repairs and maintenance and more time at high capacity. Before the use of HONE, terminals were set up in a way that forced RTGs to move to trucks for deliveries: this meant that RTGs would move containers back and forth in half-mile long rows. This inefficient method brought about a reservation system so that instead of the RTG going to the truck, the truck would go to the RTG for delivery. This showed that the reservation system was used to reduce RTG movement. As trucks enter the gate, they move to a reservation area and are then tracked by an algorithm to determine the correct time for them to line up for delivery.

 

Table 2 Average number of feet traveled per RTG in a day

 

Prior to the implementation of the HONE engine, RTGs moved an average of nearly 29,370 feet per day. This means that when the reservation system was utilized correctly, it significantly reduced the distance the RTG moved during the day; the implementation of the reservation system and delivery algorithm showed an immediate improvement in RTG movement.

 

Each row is approximately 2,200 feet long. The optimal RTG delivery pattern is to move to one end of the row on the first shift of the day, and then return to that row on the second shift. This means that in a perfect scenario, the RTG would move approximately 4,400 feet in one day. Before HONE, the RTG was moving more than seven times that amount. Once the HONE system was implemented, the RTG moved an average of 20,460 feet in a day. In the two years since the initial implementation, the Port has added more RTG rows while maintaining the reservation system. This further reduced the average distance per RTG to 16,050 feet. The biggest reason this distance has not been reduced further is that RTGs still frequently backtrack to retrieve the containers they pass.

 

In January 2022, SavantX introduced a new algorithm called Foresight: this algorithm uses quantum computing to determine the best path RTGs should take to deliver a fleet of containers.

 

Figure 6 Comparison of the quantum algorithm with the classical algorithm in the case of a queue with 2 to 5 trucks. For short container queues, the average distance per delivery (Ft.) is smaller using the classical algorithm.

 

Figure 7 Comparing the quantum algorithm with the classical algorithm in the case of 6 to 9 trucks in the queue. The average distance (in feet) per delivery is smaller using the quantum algorithm. For a long container queue, the average distance per delivery (feet) is smaller.

 

Initial testing of the Foresight algorithm in early spring 2022 showed that for short queues, the original algorithm performed slightly better than Foresight, but for long queues, Foresight was able to reduce the number of feet traveled per container by as much as 30%. This suggests that the Foresight algorithm should be used during periods of long container queueing times. Based on the analysis, if Foresight is used at Terminal 300 in July 2022, the average daily distance traveled by RTG for that month could be reduced to an average of 11,400 feet.

 

06Based on the Leap™ Quantum Cloud service, the HONE project will improve the global supply chain

 

HONE is a system that is constantly adapting and evolving, seeking to implement cutting-edge technologies to solve supply chain problems. Leveraging D-Wave's Leap™ Quantum Cloud Service, SavantX is a leader in developing and delivering quantum computing systems, software and services that improve efficiency at the Port of Los Angeles. d-Wave computers are quantum annealers designed to solve complex optimization problems such as the logistics challenges described in this white paper. By converting the problem at hand into what is known as Quadratic Unconstrained Binary Optimization (QUBO), SavantX can repeatedly utilize the D-Wave quantum computer in real time in transportation applications.

 

In the 2 years of use at Pier 300, HONE has doubled the throughput while reducing RTG movements by almost half. Trucks now use a custom reservation system to access Pier 300 in an organized and controlled manner. Quantum computing has demonstrated how it can make the terminal more efficient than ever before, increasing the value of Pier 300 to more than $2 billion after implementing HONE.

 

In the future, other HONE projects will extend quantum computing and machine learning to further improve the global supply chain.

 

About SavantX.

 

Founded in 2015, SavantX's predecessor technology has been successfully deployed at the Defense Intelligence Agency for several years. After building an optimized data analytics toolkit using large volumes of nuclear plant data, patents were filed and the company's core products are now available through a variety of services.

 

SavantX's mission is to make data easily accessible and understandable through quantum computing, intelligent search, augmented intelligence and data visualization.

 

Full report：

https://media-exp1.licdn.com/dms/document/C561FAQEn5VGU2gHUdQ/feedshare-document-pdf-analyzed/0/1668788500103?e=1670457600&v=beta&t=SrOyQOTROHHCIiHm1sWrpw9FIBtFpv2-fXUOzKwvHpw