In this section, the data collection and the analysis method will be discussed. Therefore, in Section 5.1.1 the method of data collection will be discussed and the corresponding issue of deterministic and stochastic parameters. Subsequently, in Section 5.1.2 the analysis design will be discussed. Finally in Section 5.1.3, the parameter settings for the average case will be given.
5.1.1 Data Collection
To be able to use this model a variety of input parameters is required. As can be concluded from the model description in Appendix G, information is needed about barging, trucking and handling capacities, costs, operating times et cetera. This information is gathered from the field and a previous study by Van Rooy (2010). Since this project is executed commissioned by Brabant Intermodal B.V., the willingness to share sensitive information was present. This facilitated an accurate representation of the various input parameters by asking questions during the meetings with company representatives, during the management team meetings and by asking them to fill in some questionnaires.
Most of the input parameters are deterministic and are estimated by the various terminal operators. However four parameters are stochastic and deserve extra attention when executing the numerical analysis. The first parameter is related to the natural demand growth. Second of all, two of these parameters are related to the effect of reliability on the demand, namely the percentage of growth due to an improved reliability and the reduction in rush orders due to an improved reliability. Finally, the fourth parameter is related to the reduction of the waiting times in the port as a consequence of the agreements on fixed time windows.
5.1.2 Analysis Design
There are two categories of results; the quantification of the benefits and the allocation of the value. A two step approach is used for the analysis of these results, namely:
1. Analyze an average case
The first part of the analysis is focused on discussing the results of the average case. In this average case, the stochastic parameters are set on their expected value. Thereby, the expected value is the value of the parameter as is expected by company representatives or the value of the parameter as is expected based on literature. These expected values will be discussed in Section 5.1.3. For this average case, the values for the various coalitions are determined by the use of the method as described in Chapter 3. The results of this average case will be analyzed first in Section 5.2, by studying the structure of the game, checking whether it fulfills some nice properties and analyzing in more detail some remarkable values. Subsequently, the value will be divided based on the selected allocation rule (Chapter 4) and it will be checked whether the allocation vector is in the core (Section 4.1.2.1). Finally, the meaning of this allocation in practice will be discussed. For all of the results given in this report, it needs to be noted that although they are given with a high accuracy, this accuracy is questionable due to some more inaccurate input parameters and some assumptions. However, it is decided to give these values with this accuracy, since they are used for further calculations on the division of the value.
33 2. Scenario analysis
Since some of the parameter settings in the average case are stochastic as discussed in Section 5.1.1, a scenario analysis is executed. This is done by changing one by one the values of these parameters and by studying the effect on the value (operational cost savings + extra revenues) of cooperation as well as on the allocation of this value among the partners. Furthermore, the properties of the games and the properties of the allocation vectors are checked. The results of this scenario analysis are given in Section 5.3.
5.1.3 The Average Case Settings
In this section, the settings for the stochastic parameters in the various scenarios will be discussed and explained.
5.1.3.1 The Expected Natural Growth
Ecorys (2010) studied the expected growth for the container transport. In their study is stated that a light to a very sharply increase is expected for container transport by barge. Thereby, they assume two scenarios. One in which it is expected that the growth will be 3% per year until 2040 (represented as the SE scenario). And another, more optimistic, in which they state that the growth will be 6 % per year until 2020 and from then on 4% until 2040. The trend line based on the historical information results in a scenario in between those scenarios. These scenarios are illustrated in Figure 5. Based on their verification of these expectations, they concluded that the future development will be above the trend line. This is concluded since a demand growth is expected due to the growing international trade and the repositioning of production to Asia.
However, besides this change, some other developments need to be considered like; the modal split agreements for port terminals of Maasvlakte 2, the ever-increasing containerization of goods, the increase in costs for trucking due to the road pricing, more attention for sustainable developments within the logistic supply chains.
Figure 5: The expected demand growth of container transport based on three scenarios. (Source; Ecorys based on CBS, CPB/DVS.
Furthermore, they make a distinction in the expectations for the various regions. They state that the South of the Netherlands, including Noord-Brabant, expects a strong growth directed towards the GE scenario as given in Figure 5. Based on these findings, it is decided to assume an average growth for the next year of 5 %.
34
5.1.3.2 The expected increase due to a more reliable product
Besides this natural growth, it is expected that the volumes of the partners within Brabant Intermodal B.V. grow as a consequence of a more reliable service that can be offered due to cooperation. The effects of cooperation on the reliability of the service and subsequently the effects of reliability of the product on demand are stochastic. Therefore, a reasonable estimation needs to be made for the expected growth. Based on discussions with company representatives it is decided to set this expected growth when cooperating in the grand coalition on 3% and without cooperation on 0%. To be able to use game theory, also expectations need to be made for cooperating in other coalitions. It is assumed that this growth is increasing, which means that the growth becomes relatively larger when the coalitions become larger. For the average case scenario, the percentages are set for the single-player, the two-player and the three-player coalitions respectively on 0, 0.5% and 1.5%. This growth is assumed to be equal for each terminal in the various coalitions.
5.1.3.3 The expected percentage of rush orders
Besides the growth of the volume, it is expected that a more reliable product decreases the number of rush orders. This is assumed since the customer will get more confidence in a timely delivery by intermodal transport. Also this effect is uncertain, since cooperating and bundling can have a negative as well as a positive effect on the number of rush orders due to the longer throughput time in contrast to a higher reliability that is expected. In all scenarios, the percentages of rush orders are assumed to remain equal to the current percentages for the single coalitions. This means for ITV, ROCW, BTT and OCT respectively 6%, 5%, 30% and 5%.
Based on discussions with company representatives, it is decided to set the expected percentage of rush orders for the grand coalition below the current percentage of rush orders, which means respectively for ITV, ROCW, BTT and OCT at 3%, 2%, 24% and 2%. From these percentages can be concluded that the terminal with a high level of rush orders has a stronger percentage decrease, than the terminals with low levels of rush orders. Furthermore, for the various coalitions is assumed that the decline is increasing. Hence, a larger coalition has a relatively larger effect on the reduction of rush orders than the smaller coalitions. The percentages used in the average case scenario are given in Appendix R.
5.1.3.4 The waiting times at the port terminals
In accordance with Konings (2009) attention has to be shifted towards improving the quality of information exchange between actors, the introduction of fixed time windows and better route planning of terminal visits to reduce the waiting times at the port terminals. In accordance to Van Rooy (2010) the average waiting time at port terminals is related to the number of calls that is made in the port. This is resembled in Table 7, which represents the relationship between the number of terminal calls and the total waiting time in the port terminals of ECT cumulated for barges operating connections between the port and eight Dutch inland terminals. The first column in this table represents the number of terminal calls made during one rotation. In the second column the number of occurrences of the specific number of terminal calls during one rotation is summed up. The third column gives the accumulated waiting time in the port for the whole rotation, with in the fourth column a corresponding standard deviation. Finally, the average waiting time per call is given in the last column. Based on this table a weighted average waiting time per call is determined. This weighted average waiting time is equal to 2.1 hours. A critical note made on this calculation is that for calculating the values in the table below data was used of the year 2009. The trade volumes and the utilization rate of the port terminals were low during this year due to the economic crisis. Since the utilization rate in the port is increased, the waiting time is in the average case assumed to be 2.5 hours. Although the data in Table 7 are based on the port terminals of ECT, it is assumed that these waiting times are equal for the other terminals included in this project (Euromax, APMT and the Empty Depots).
35
Table 7: Average waiting times at the port terminals and the dependency on the number of terminal calls (Van Rooy, 2010).
The best way to reduce these waiting times is in accordance to inland terminal operators the use of fixed windows for barge handling at the port terminal. A fixed time window means that a certain amount of quay capacity is reserved for one specific barge for a specific time period (Van Rooy, 2010). Douma (2008) has proven that fixed time windows reduce waiting times in a cooperative environment. Since there is no information about the waiting time reduction under fixed time windows a reasonable assumption needs to be made. It is assumed that this average waiting time will become 0.5 hours. This is a relatively low, but feasible waiting time, since in a fixed time window it is agreed that the quay is always available at a specific time. However, it needs to be mentioned that this waiting time is only feasible under the assumption that the barges of the cooperation are always on time. This means that they are not earlier or later than the agreed time window. In practice, the throughput time of a barge is unpredictable and hence a barge can arrive later or earlier than the agreed time window, which can result in longer waiting times than assumed in this expected value.