SECTOR DE LAS ARTES PLÁSTICAS Y LOS CENTROS MUSEÍSTICOS

Having field tested the pushback rate control protocol, the next step is to quantify the benefits of the approach. The main dimensions of the benefits that we address are the taxi-out time and fuel burn reductions. Intuitively, it is reasonable to use the gate-hold times as a surrogate for the taxi-out time reduction, as long as runway throughput is maintained. We test this hypothesis through a simulation of operations with and without metering.

Simulation set-up

The purpose of these simulations is to estimate the taxi-time savings and to investigate the fairness of the strategy in terms of the distribution of gateholds. In particular, we compare three different sets of outcomes:

1. Data from actual operations: This case corresponds to the system behavior during the push-back rate control demo periods. The taxi-times and queuing times are measured using ASDE-X data.

2. Simulation predictions: This case corresponds to the simulated output of pushback rate control demo periods. In this simulation, flights are cleared for pushback at the same times that they received pushback clearance (after being assigned gateholds) during the demo.

3. Hypothetical (no pushback rate control) simulation: Finally, the model is used to simulate what would have happened if pushback rate control was not in effect, that is, if flights had been cleared for pushback as soon as they called ready to push. In the simulations, the pushback clearance times for flights are set to be equal to the call-ready times, that is, all gate-hold times are set to zero.

The common elements in all simulations are the following:

1. The departure slots are fixed and determined by the data from actual operations for each day. This reflects the fact that there are differences in runway performance across days due to factors not related to the pushback rate control strategy.

2. The flights with EDCTs and DSPs are assumed to have fixed departure times-same as the ones observed in real operations. This is because these flights have pre-defined departure times.

The first step is to determine the unimpeded taxi-out times of flights using ASDE-X data, adopting the procedure outlined in Section 4.3.1. Given the pushback clearance time, the unim-peded taxi-out time and a taxiway congestion component, each flight is propagated to the runway, where it is assigned to the next available departure slot for that time period, which determines the predicted wheels-off time. The difference between this wheels-off time and the pushback clearance time is the expected taxi-out time.

The fixed departure slots are a reasonable assumption as long as there is a nonzero queue at the departure threshold. The total and mean taxi-out times from the actual data and the model predictions are expected to be the same, since the pushback times and departure slots are the same for both cases. The additional comparison of the actual and predicted runway queuing times would reflect the ability to predict the travel time from the ramp to the runway queue, and subsequently to compare the impact of the control strategy using the simulations.

The results are summarized in Table 5.3 for the two days with significant gateholds. The results pertain to flights that were released for pushback between 1675 and 2045 hours, that is, near and during the metering period. There were 21 flights with EDCTs and DSPs on July 21 and 17 such flights on July 22. As can be seen in Table 5.3, the mean taxi-out time and the mean queuing time (the time an aircraft spends in the departure runway queue) are generally predicted very well by the model.

Table 5.3: Effect of gate-holding on mean taxi-times and queue lengths.

Date

Actual operations Model predictions No pushback rate ctrl.

No. of Taxi-out Queuing Gate-hold Taxi-out Queuing Taxi-out Queuing flights time (min) time (min) time (min) time (min) time (min) time (min) time (min)

7/21 121 16.5 5.7 368 16.5 5.8 19.5 7.9

7/22 121 17.9 7.2 279 17.9 7.4 20.2 9.2

In the top part of Figure 5-10 we show the instant actual and simulated queue on July 21. They match very well, so the actual queue is predicted accurately by the simulations. In the bottom part of the same figure we compare the simulated queues of July 21 with and without PRC. The evident difference between the simulated queue sizes shows the benefit of the pushback rate control strategy.

17 18 19 20 21 0

2 4 6 8 10 12

Number of jets

Local time Jul21Actual Queue Jul21What−if Queue

17 18 19 20 21

0 2 4 6 8 10 12

Number of jets

Number of jet aircraft in queue

Jul21 Actual Queue Jul21 Simulated Queue

Figure 5-10: Queue sizes measured and predicted per minute, on July 21, 2011.

In addition, we conduct a benefits analysis of the fuel burn savings by using the simulated taxi-out time savings times as a first-order estimate of the actual taxi-out time savings using the methodology outlined in prior work [70, 110]. The total fuel savings are estimated to be 2,650 US gallons, which translates to average fuel savings per gate-held flight of about 57 kg.

Distribution of benefits

Equity is an important factor in evaluating potential congestion management or metering strategies.

The PRC approach, as implemented in these field tests, invokes a First-Come-First-Serve (FCFS) policy in clearing flights for pushback. One would therefore expect that there would be no bias toward any airline with regard to gateholds incurred, and that the number of gateholds for a particular airline would be commensurate with the contribution of that airline to the departure traffic during the congested periods. However the taxi-out time saving predicted by the simulations is not equal to the gate holding time of each individual flight. Thus, the taxi time savings of each carrier can differ from the total time flights of this carrier were held at the gate as can be seen in Figure 5-11. This is because the benefit of holding a flight at the gate can spill over to other flights as well as explained in earlier studies of N-Control [107]. In short there are two main reasons for

this:

• Overtaking: Consider a scenario in which aircraft A calls for push and is authorized to push. Aircraft B calls for push just a few seconds later and is held at the gate for 3 minutes.

Subsequently, aircraft B pushes back and finds itself in the departure queue behind aircraft A. In absence of the metering program, aircraft B would have pushed a few seconds and not 3 minutes later than aircraft A. This might have been enough time for aircraft B to overtake aircraft A if it taxied at a faster speed, its gate was located closer to the runway, or it performed the pushback process faster. Thus, in the counterfactual scenario, aircraft B could have ended up in front of aircraft A in the departure queue. In such a scenario the cost and the benefit of the program could swap between two flights.

• Re-scheduling despite same sequencing: Consider a scenario in which all pushback slots have been utilized and there are 5 more minutes left until the end of the current time-window. In this time-window 4 new flights call for pushback, each one a minute apart from the previous one. When the next time window commences, the rate is set to 3 every 5 minutes.

The controller authorizes the first 3 aircraft to push back together, and the forth one five minutes later.

As a result of these two phenomena and their combinations, the benefits of the metering scheme can be divided and allocated between flights in an unpredictable manner. However, the first come first serve sequence is maintained and in general the gateholding times would be approximately equal to the taxi-time reduction experienced by each airline, as can be confirmed from Figure 5-11. However, the actual fuel burn benefit also depends on its fleet mix. Figure 5-11 shows that while the taxi-out time reductions are similar to the gateholds, some airlines (for example, the ones denoted Airlines 4, 13, 21 and 27) benefit from a greater proportion of fuel savings. These airlines are typically ones with several Heavy aircraft during the evening times.