PORTS PCP: SIMULATING THE SHIP-ANCHORAGE CONNECTION TO REDUCE DEMURRAGE COSTS – PART 2

For each of the combinations of berth allocation rules and queue priorities, 30 replicates of 2007 days of operations were performed, and queue waiting time statistics were collected for the system as a whole and for each of the ships that periodically dock in port. It is worth mentioning, as a methodological observation, that the SBL (ship-berth link) model focused on ships that visit the port at least twice a year. In 14, 23 ships out of XNUMX met this criterion. They constituted the object of analysis in this case study.

Figure 2 shows a screen of the main SBL models developed in Arena.

Figure 2: A screen from one of the models (dedicated berths)

The control variables are the average interarrival times and the average berth processing times for each ship. It was also assumed that port resources (equipment, capacities, and so on), beyond the scope of the modeled SBL operation, did not influence the statistics collected for time spent in queue per vessel.

The interarrival times for each ship were assumed to be exponentially distributed. According to Dragovic et al. (2005), the distribution of time between arrivals is a basic contributing parameter that must be assumed or inferred from observed data. The most commonly assumed distributions in the literature are the exponential distribution (Demirci, 2003; Pachakis and Kiremidjian, 2003); the negative exponential distribution (Shabayek and Yeung, 2002) and the Weibull distribution (Tahar and Hussain, 2000).

It was also assumed that berth processing times for each vessel were normally distributed. Four sublevels of the coefficient of variation for processing times were tested in each combination: 0; 0,1; 0,2; and 0,4. Although the total ship processing time depends not only on the number of containers lifted, but also on the number of wharf cranes allocated per ship, a simplification is made here for the purposes of this case study. Issues related to the number of lifts and quay cranes were therefore considered to be integrated into these average processing times for the sake of simplification.

RESULTS

Figure 3 presents the expected values ​​of the average waiting time in the queue for each ship in each of the combinations between the berth allocation norms and the queue priorities. It is clear that these opposing forces within the berth allocation norms and queue priorities for each ship must be taken into account together. More precisely, they must be weighted not only by the number of arrivals per year, but also by the demurrage costs of each ship, to determine which combination would actually lead to the lowest total demurrage cost for the entire system.

Description:1 – Dedicated berths per vessel type (size – large/small) 2 – Single file delivers ships to the first available berth 3 – Ships allocated to the anchorage with the smallest queue size 4 – Ships allocated to an anchorage with shorter queue time a – Longer service time b – Shorter service time c – Highest number of visits per year d – Lower number of visits per year and – PEPS f - UEPS g – Largest ship size h – Smallest ship size

In a more detailed view of Figure 3 and observing the characteristics of each ship, it is possible to state that small ships would benefit more from rules in which berths are dedicated by ship size or a single queue distributes ships to the first available berth . Small ships are also less affected by the coefficient of variation of processing times. With regard to queue priorities, there appears to be a trade-off between the shortest processing time and the smallest ship size in terms of their impact on average queuing time.

On the other hand, when considering large ships, it can be seen that they would benefit more from rules where a single queue distributes ships to the first available berth or ships are allocated to a berth with the shortest queue time. With regard to assigning a specific queue priority, however, the result is not so clear. It is also possible to state that, unlike small vessels, large vessels are more affected by the coefficient of variation of processing times.

Different levels of critical waiting time (Wq_critical) were simultaneously tested with different levels of demurrage cost coefficients. The demurrage cost coefficient indicates how many times this cost per hour is higher on a large ship than on a small one. The total demurrage cost was calculated for each of the combinations of berth allocation norms and queuing disciplines. The best combination for each pair of Wq_critical and overstaying cost coefficient was then identified and plotted in Figure 4.

From Figure 4, it can be seen that when the Wq_critical is low (that is, the waiting time spent in queues is a critical component of the services provided by the ports) and the overstaying cost coefficient is also low ( i.e. the cost of overstaying for a small ship is higher than for a large ship), port authorities should allocate ships to the berth with the shortest queue time and assign priority to the ship with the shortest queuing time. processing. In this case, clear priority would be given to small ships, in order to favor the performance of the port system as a whole. Ships #5 and #8 in the Figure, for example, would be exceptions to this logic, as they are large ships with short processing times.

As the demurrage cost coefficient increases, but the Wq_critical is still low, port authorities should adopt a single queue to distribute ships to the first available berth. Priority, however, should be given to ships with the highest number of visits per year. According to this priority criterion, large ships with large numbers of visits per year would be favored, thus helping to reduce the system's total demurrage cost. Notices #6, 7, and 8, for example, meet this criteria.

Finally, as the Wq_critical increases and, as a direct consequence, service standards in the port become less rigorous, the system quickly moves to the classic PEPS discipline, with ships being allocated to the anchorage with the shortest queue time at the time of their arrival. arrival. There seems to be a compromise between the Wq_critical level and the overstay cost coefficient. The larger the first, the smaller the impact of the overstay cost coefficient in determining the most appropriate standard combination.

CONCLUSION

The SBL problem is complex, due to different ship arrival times, different ship sizes, different berth processing times, and so on. Through a simulation of a small port with two berths, this case study evaluated the impact of different berth allocation norms and queuing priorities on waiting time, both in terms of their values ​​and expected variances. Their findings represent a contribution not only to theory but also to practice in decision-making. Three main elements, detailed below, constitute the contribution to this study in terms of previous literature.

The first element is the experimental confirmation, through simulation, of the available quantitative evidence found in the literature list on assigning queue priority to ships with shorter processing time, in order to improve the overall performance of the port. Processing time is often positively correlated with vessel size and negatively correlated with its number of visits per year, two facets of the same operational characteristic.

The second element is the identification that different types of ships are differently affected, not only by different combinations of berth allocation norms and queuing priorities, but also by different levels of the coefficient of variation of processing times, thus indicating that it is necessary to take into account their specificities, such as costs of overstaying, number of visits per year and variances in waiting time.

Finally, the third element is the experimental indication of the apparent contradictions or compromises between the analysis of the SBL operation as a whole and the SBL operation as a weighted sum of the expected demurrage time of each ship, by its number of visits per year and for your demurrage cost per hour.

With regard to this last element and its practical aspect in decision-making, the main issue found by the study is the great importance of considering the specificities of each ship for an adequate positioning of the port authorities in terms of rules for the allocation of berths and queue priorities. This explains why the simulation results, generated for each vessel, still have to be further analyzed and weighted in terms of overstay probabilities and costs.

BIBLIOGRAPHY

Demirci, E.. 2003: Simulation modeling and analysis of a port investment. Simulation 79: 94–105.

Dragovic, B.; Park, NK; Radmilovic, Z.; Maras, V.. 2005: Simulation modeling of ship-berth link with priority service. Maritime Economics & Logistics 7: 316-355.

Pachakis, D.; Kiremidjian, AS. 2003: Ship traffic modeling methodology for ports. Journal of Waterway, Port, Coastal and Ocean Engineering, ASCE 129: 193–202.

Shabayek, AA; Yeung, WW. 2002: A simulation model for the Kwai Chung container terminal in Hong Kong. European Journal of Operational Research 140: 1–11.

Tahar, MR; Hussain, K.. 2000: Simulation and analysis for the Kelang Container Terminal operations. Logistics Information Management 13: 14–20.