IV. Numerical Experiment
4.3 Result Analysis
4.3.2 Large-scale Network Results
Scenario 1: Hong Kong International Airport failed
The results of the scenario which is failed in Hong Kong Airport are given in following figures and tables.
We show the alternative routes when the disruption occurs with different severe level in Table 4.17. The shipments in the scenario 𝑟18= 0.75 to 0.2 have little changes in the routes. By and large they will choose the new alternative route from Hong Kong (HKG), Dubai (DXB) to Leipzig (LEJ) or remain to be transported by original route which is from Hong Kong (HKG) to Leipzig (LEJ). This is because the failed HKG limits the links capacities from HKG to the other airports. It makes the cargos transported by different routes. In the scenario 𝑟18 = 0.15 and 0.1, the rail transportation is chosen in Japan because the cost is lower than renting the trucks and the sum of travel time and renting time is just a little higher than renting the trucks.
When the HKG hub remains 10% capacity or less, the sea transportation instead of air is used between Taiwan to China. The shipment will depart Kaohsiung port (KHH) or Keelung port (KEL) to Shanghai port (SHA).Under these scenarios ( 𝑟18 = 0.1~0 ), Hong Kong Airport (HKG) can’t provide the enough capacities. When Hong Kong Airport (HKG) totally can’t work, shipment 1 will not be transported and the other shipments are consolidated in Shanghai Pudong Airport (SHA), and then delivered to the Leipzig/Halle Airport (LEJ) by air directly. The trucks are still responsible for the inland transport in Europe.
We find most of shipment will be transported to the destinations before the cargo time value declines. Due to the high values of the technologic products (short-life cycle
commodities with high values), the shipment 2 is always be transported with the entire amounts and on time no matter the disruption is serious or not, except for the scenario 𝑟18 = 0.1. In the scenario 𝑟18 = 0.1, the shipping time of shipment 2 (technologic
products) is over 5760 minutes, so the cargo time value of technologic products decline to 13670000 dollars per ton. When the disruption becomes more severe, part of cargos in the
Table 4.17 The alternative routes for HKG hub failure (without the resilient strategies)
3357.75
OSC →HNDHKGDXB LEJ→ Berlin Zoh → TPE HKG DXB LEJ→ Greven
OSC →KIX HKGDXB LEJ→ Berlin Zoh → TPE HKG DXB LEJ→ Greven
OSC →HND HKGDXB LEJ→ Berlin Zoh → KHH… SHA LEJ→ Greven
shipment 1 and 3 are abandoned because it needs to consider the resilient cost, too. When Hong Kong Airport (HKG) totally can’t work, the constant value products (shipment 1) are abandoned totally so that the company can rescue more holiday gifts (shipment 2) with the less cost spending.
The renting activities for HKG hub failure scenario are listed in Table 4.18. Most of trucks are rented from carrier 1 because of the lower renting cost. There is an exception in the scenario 𝑟18= 0.075, it rents one ton trucks capacities from carrier 2. We can infer that when the rented capacities are few, it may rent the trucks from the carrier with shorter renting time to save the time spending instead of saving cost. For the air and sea transport, we find that most of aircraft containers are rented from non-partners because the
contractual partner doesn’t accommodate the capacities for the requisite routes. The same situation also happens to the sea transportation. Thus, the selection for the air and sea carriers is limited by the airlines and shipping lines which set up from contractual partners.
Table 4.18 Renting capacities for HKG hub failure Capacity
remaining proportion (Severe level)
Link mode carrier rent amounts
(ton)
𝒓𝟏𝟖 = 𝟎. 𝟕𝟓 (OAA, KIX) truck carrier 1 3
𝒓𝟏𝟖 = 𝟎. 𝟐 (OAA, KIX) truck carrier 1 3
𝒓𝟏𝟖 = 𝟎. 𝟏𝟓 (OAA, NGO) rail Japanese National Railways
26
𝒓𝟏𝟖 = 𝟎. 𝟏 (LEJ, Hanau) truck carrier 1 3
(OAA, NGO) rail Japanese National Railways
27
(KHH, SHA) ship Evergreen 20
𝒓𝟏𝟖 = 𝟎. 𝟎𝟕𝟓 (OAA, KIX) truck carrier 2 1 (KEL, SHA) ship carrier 1
(Evergreen)
20
𝒓𝟏𝟖 = 𝟎. 𝟎𝟓 (KEL, SHA) ship carrier 1 (Evergreen)
20
𝒓𝟏𝟖 = 𝟎. 𝟎𝟐𝟓 (KEL, SHA) ship carrier 1 (Evergreen)
20
𝒓𝟏𝟖 = 𝟎 (OAA, KIX) truck carrier 1 2
(KIX, SHA) aircraft carrier 2 (non-partner)
24
(KEL, SHA) ship carrier 1 (Evergreen)
20
The changing allocation of own trucks (Table 4.19) happens at the service center in Japan primarily because the location of the disruption is in Japan. When the disruption becomes more serious, it spends the recovery cost on the Japan service centers rather than on the Germany service centers. Thus, we know that the model will recover the more critical parts first. In general, the trucks re-allocations are changed depending on the route choice.
In Table 4.20, we count the amount of own trucks re-allocation and renting capacities for failed HKG hub. In the slight disruption, it does the trucks re-allocations and rents the trucks.
Sometimes considering to arrival time, the Japan rail is chosen to reduce the shipping time.
When the capacities remain 0.1, it starts to rent the ship containers and when the HKG hub totally can’t work, 24 tons capacities for aircraft containers will be rented. In general, the amounts go up gradually. It means when the situation becomes more serious, the recovery activities will increase.
Table 4.19 own trucks re-allocation at nodes for HKG hub failure Scenario
(Severe level)
node rent amounts (ton)
𝒓𝟏𝟖 = 𝟎. 𝟕𝟓 OAA
Table 4.20 The amounts of own trucks re-allocation and renting capacities for failed
In order to proof the effectiveness of our model, we compare our first objective value with the value without resilient strategies which is obtained from the revised model. We delete the decision variables including re-allocating trucks and renting capacities from our resilient model and change the objective functions to be maximizing total throughputs and minimizing the travel time in the revised model. The comparing results are list in Table 4.21 and show in Figure 4.4. We find the increased percentage of cargo time value enhances increasingly when the disruption gets more serious.
Furthermore, the recovery cost raises by the situation getting serious, but when the capacities remain 5% of HKG hub, the cargo time value starts to decline and recovery cost reduces. It can be interpreted that the model considers the two of objective functions and decide to give up some cargos to reduce the recovery cost. When the Hong Kong Airport totally doesn’t work, the express company needs to pay a lot of money to recover the cargo time value from 0 to 516000 thousand dollars. Thus, it is worth noting that if we want to recover from very severe disruption, the express company needs to pay a lot or it may implement the proactive resilient strategies to prevent its disruption.
Table 4.21 Cargo time value and recovery time for HKG hub failure Capacity
remaining proportion (Severe level)
without resilient
strategies with resilient strategies cargo time value
(thousand dollars)
cargo time value (thousand dollars)
increased percentages
recovery cost (dollars)
𝑟18= 0.75 500000 543000 8.6% 13479.5
𝑟18= 0.2 500000 543000 8.6% 14179.5
𝑟18= 0.15 500000 551000 10.2% 148900
𝑟18= 0.1 408000 562000 37.75% 207746
𝑟18 = 0.075 275000 490000 78.2% 19727.5
𝑟18= 0.05 178000 418000 134.83% 2700
𝑟18= 0.025 81000 357000 340.74% 2000
𝑟18= 0 0 516000 1476381
Figure 4.4 Cargo time value and recovery cost for HKG hub failure
Scenario 2 : Japan Earthquake
In the Japan earthquake scenario, we assume the combination of link, node and mode failed situation. The resulting new alternative paths are shown in Table 4.22.
Because Narita International Airport (NRT) is closed and the links from Tokyo International Airport (HND) to HKG and SHA leave few capacities, the cargos are transported from service centers in Japan to Kansai International Airport (KIX) and Central Japan International Airport (NGO). The constant value commodities and holiday gifts will go through Hong Kong International Airport (HKG), Dubai International Airport (DXB) and Leipzig/Halle Airport (LEJ). The technologic products (short life-cycle commodities) are transported from Shanghai Pudong Airport (SHA) to Leipzig/Halle Airport (LEJ) directly. Otherwise, due to the road system and rail are broken in Japan, only part of cargos can be delivered.
Comparing with the results from the model without resilient strategies, our model recovers the cargo flows of 14 ton technologic products (short life-cycle commodities) from Nerima (TYN) to Greven (Grv). On the other hand, it is worth noting that the results in our model may not choose the paths with shortest travel time, for example shipment 1 and 2, but it must choose the paths that allow the shipments be delivered before the cargo time value declines and the paths which will not have additional expenditure.
Table 4.22 The alternative routes for Japan Earthquake in the two situations
→ truck - - - rail air … sea
OSC→ KIXHKG LEJ→ Bel NGC→ NGO HKG LEJ→ Han
The renting activities for Japan earthquake scenario are listed in Table 4.23. We find most of renting capacities are used to transport the cargos of shipment 2 including the trucks rented from Nerima (TYN) to Kansai International Airport (KIX) and the aircraft containers rented from Kansai International Airport (KIX) to Shanghai Pudong Airport (SHA). But there is an exception which is the trucks rented from Nagoya Central (NGC) service center to Central Japan International Airport (NGO). It is because the trucks get broken in this scenario.
No matter the shipment is from Nerima service canter (TYN) to Kansai International Airport (KIX) or from Nagoya Central service center (NGC) to Central Japan International Airport (NGO), it rents the trucks from carrier 1 because the renting cost is cheaper than carrier 2. Until the capacities of carrier 1 are exhausted, the capacities of carrier 2 are rented. It shows that the time spending on dispatching the trucks (𝑅𝑇𝑖𝑚𝑒) by different carriers is not the significant factor which impacts the renting decisions.
The reason why aircraft containers are rented from non-partner is that the partner carrier doesn’t set up the route from Kansai International Airport (KIX) to Shanghai Pudong Airport (SHA).
Table 4.23 Renting capacities for Japan Earthquake
Link mode carrier rent amounts
(ton)
(TYN, KIX) truck carrier 1 10
truck carrier 2 1
(NGC, NGO) truck carrier 1 3
truck carrier 2 1
(KIX, SHA) aircraft carrier 2 (non-partner)
14
The changing allocation of own trucks happens at Nagoya central service center
(NGC). Five trucks (5 ton.) are re-allocated to transport the cargos from Nagoya central service center (NGC) to Central Japan International Airport (NGO).
Scenario 3 : Icelandic volcano eruptions
In the Icelandic volcano eruptions scenario, we assume that all the air links from the other airports to Europe airports are all failed. The resulting new alternative paths are shown in Table 4.24. It is noted that intermodal transportation, which is integrated with air and rail, becomes the best way to transport the cargos from Asia to Europe. They are transported from Hong Kong International Airport (HKG) or Shanghai Pudong Airport (SHA) to Moscow by air, and then transported to Parchim International Airport (SZW) or Leipzig/Halle Airport (LEJ) by rail.
We find that shipment 2 (technologic products with high value) and Shipment 3 (holiday gifts) are transported to the destination before the cargo time value declines. In order to maintain more cargo time value, the model lets the technologic products and holiday gifts be transported by the routes with shorter travel time. Besides, comparing to the results of the model without resilient strategies, our model recovers the amount of delivered cargos from 0 to 48 ton.
→ truck - - - rail air … sea
Table 4.24 The alternative routes for Icelandic volcano eruptions The results
The renting activities for Icelandic volcano eruptions scenario are listed in Table 4.25. It rents plenty amount of aircraft containers and rail capacities. Most of aircraft containers are rented from non-partners because the contracted partner doesn’t set up these routes.
Table 4.25 Renting capacities for Icelandic volcano eruptions
Link mode carrier rent amounts
(ton) (NRT,SHA) aircraft carrier 2
(non-partner)
20
(NGO, SHA) aircraft carrier 2 (non-partner)
8
(HKG, MOW) aircraft carrier 2 (non-partner)
20
(SHA, MOW) aircraft carrier 2 (non-partner)
8
(TYN, NRT) rail Japanese National Railways
20
(MOW, LEJ) rail 8
(MOW, SZW) rail 40
The changing allocation of own trucks happens in the Osaka Central service center (OSC) and at Parchim International Airport (SZW). Ten trucks (10 ton.) are re-allocated to Osaka Central service center (OSC) to transport the cargos from OSC to Kansai International Airport (KIX). One truck (1 ton.) is re-allocated to Parchim International Airport (SZW) to service the route from SZW to Greven (Grv).
The comparison between all the scenarios
In the following Table 4.26, we know that different scenarios have different renting activities. It is worth noting that it rents large quantities of the rail and aircraft containers in the Icelandic volcano eruption.
Table 4.26 The amounts of own trucks re-allocation and renting capacities for all scenarios
To confirm the effectiveness of our model in the different scenarios, the total cargo time value with the resilient strategies and without the resilient strategies are shown in Figure 4.5 and Table 4.27. The results show that our model which considers the several recovery activities can lead to significant improvement in the total cargo time value. It indicates that the recovery activities play the important role in the aftermath of a disruption. It worth noting that the resilient cost spending on volcano eruption is much higher than the other scenarios. It is the most severe disruption to the express company.
The company may need to focus on the proactive resilient strategies in advanced.
Table 4.27 Total cargo time value and incremental recovery cost for different scenarios Scenario without resilient
strategies with resilient strategies cargo time value
(thousand dollars)
cargo time value (thousand dollars)
increased percentages
recovery cost (dollars) Japan
Earthquake
172000 382000 122% 1344937
Icelandic volcano eruptions
0 392000 6190646
The failure in HKG hub
𝑟18= 0
0 516000 1476381
Figure 4.5 Total cargo time value and incremental recovery cost for different scenarios