Lecture_W10

Page 3 Modelling freight is an important input and requirement for a range of planning and assessment projects. Freight models assist us to: • better understand and estimate the scale of freight transport, and its geographical and temporal aspects • estimate the impact of a range of policy measures and projects • estimate the economic, social, and environmental impacts

Page 6 Scope of Freight Models • Intra-urban: Freight movements made by commercial vehicles entirely within a specified urban area • Intrastate: Intrastate freight movements are made by commercial vehicles, rail, marine or air entirely within a specified state or territory. • Interstate: Interstate freight movements are made by commercial vehicles, rail, marine or air between one state or region and another. • International: International freight movements are made by marine or air between one state or region and different countries.

Page 9 1. Trip-based models It is an application of the four-step model approach that is commonly used in urban passenger demand modelling. For this type of modelling, the study area is first divided up into geographical zones that are treated as both origin and destination zones for the modelling. Trip generation : Estimation of the number of trips generated from each zone Trip distribution : Estimation of how trips originating in each zone are distributed to destination zones. At the end of the trip distribution step, the output is a trip matrix for one or more time periods, showing the number of trips between each origin- destination pair. Mode choice : Determination of mode selection. With nearly all urban freight carried on road, the mode choice step is typically not required in urban freight modelling. Trip assignment : Assignment of trips to a designated road network.

Page 12 Structure of Commodity Flow Models Source: ATAP, 2021 • Some long-haul flows can be moved entirely by commercial vehicle, moving between an external location and one or more locations within the study area. • Commodity flow models are not typically deployed solely within urban geographies and are mostly used to model interstate or intrastate flows.

Page 15 Structure of Tour-based Models • The approach begins with the generation of tours, as opposed to the generation of trips. The tour aims to replicate observed itineraries and a specific vehicle type is tied to the commodities and consignees (the stops) along the tour. • Data should be collected on the activity or purpose at each stop the nature of the commodities carried, the stop sequence, the actual routing, the industry at each stop, the activity performed at each stop, the type of vehicle used and the time of day. • Once a list of the tours and the associated vehicle type is generated, each tour can be populated with the associated intermediate stops. The formulation models the sequence of stops within the tour. • The modelling of the tour sequence and its attributes can be conducted using micro-simulation techniques, similar to the micro-simulation of households in activity-based passenger travel models. • Once the stop patterns have been generated, each leg can be transformed into a CV origin- destination, which serves as the basis for the CV matrix for trip assignment. Source: Original Calgary model developed by Hunt & Stefan, 2007)

Page 18 6. Supply chain models • The process starts by using transportation problem to model the distribution channels between shippers/customers. • We estimate commodity flows, matching consumers and producers (buyers and sellers) as the basis for distributing (allocating) flows among origins and final destinations. • From these flows, logistics chain models allocate flows by transport mode according to shipment size and frequency, available means of distribution (network supply, cost, etc.) and the availability of intermodal transfers. • The resultant shipments by mode can then be assigned to individual modal networks by vehicle routing models. • For freight models, it is the manifestation of these flows on the urban road network, including road-based trips to and from intermodal terminals.

Page 21 In summary Cost and data availability are important for selecting the type of freight model. If the data required to build a tour-based model are available, then a tour-based model is recommended rather than a trip-based model. Otherwise, commodity flow models can be considered. Source: ATAP, 2021

Page 22 Existing Freight Models in Australia Description Type of model City/location As part of the STM, there are several freight models; for heavy vehicles a commodity-based flow model is utilized to determine distribution of trucks, whilst truck trip models are used for light commercial vehicles. The key models in this suite were updated in 2017 based upon the 2014 ABS survey data, heavy vehicle traffic counts and other information provided by relevant stakeholders (ABS 2015). Further development work for light commercial vehicles (LCVs) was undertaken in March 2020. The models estimate LCVs, heavy commercial vehicles (HCVs, i.e. rigids and articulated). Commodity Flow and Trip-based Sydney The Queensland Freight Model (QFM) is a commodity flow model that extends into urban areas, but it is fundamentally a state-wide model with a more inter- regional focus. Commodity Flow Brisbane The Freight Movement Model (FMM) was developed around 2008 and a light touch recalibration was done in 2016 against the Freight Movement Survey data of 2014. The model models Rigid and articulated HCVs, whilst B- doubles and HPFVs are modelled as part of a separate port module. The whole model is incorporated into the Victorian Integrated Transport Model (VITM). Commodity Flow Melbourne

Page 23 Outline  Freight transport modelling  Supply chain resilience  Green logistics

Page 24 Supply chain Risks Definition: Risk is defined as the expected outcome of an uncertainty event, resulting in undesirable consequences. Supply chain vulnerability/risk is likely to result in a loss or damage, and can be classified into two categories: • Internal risks: such as human resources, infrastructure and equipment risks, material risk, loss of supplier, IT breakdown or technologies of machinery, … • External risks: aspects that influence demand at the level of the end customer and have a severe impact on the area of their occurrence, such as flood & bushfire, sabotage and terrorist attacks, US-China trade war, COVID, labour strikes, emerging technologies such as sharing economy providers.

Page 25 Supply Chain Risk Management Supply chain risk management aims to identify the potential sources of supply chain risk and implement appropriate actions or strategies through a coordinated approach among supply chain actors, to avoid or reduce supply chain vulnerability & risk. It is an iterative process composed by four main activities: • Risk Identification • Risk Assessment • Responding to Risk • Monitoring • Review One way to deal with supply chain risk is to increase confidence in the supply chain (i.e. confer to the supply chain the ability to be resilient) and reduce the likelihood of risk events occurring.

Page 26 Supply chain resilience Definition: The adaptive capability of the supply chain to prepare for unexpected events, respond to disruptions, and recover from them by maintaining continuity of operations at the desired level of connectedness and control over structure and function. Source: Road and Rail Supply Chain Resilience Review, Australian Government, Bureau of Infrastructure and Transport Research Economics, Feb 2023

Page 27 Some key external risks of transport network Source: Road and Rail Supply Chain Resilience Review, Australian Government, Bureau of Infrastructure and Transport Research Economics, Feb 2023

Page 28 Source: Road and Rail Supply Chain Resilience Review, Australian Government, Bureau of Infrastructure and Transport Research Economics, Feb 2023

Page 29 Quantifying the Supply Chain Resilience Resilience properties:  amount of change the system can undergo and still retain the same controls on function and structure  degree to which the system is able of self organisation  ability to build and increase the capacity for learning and adapting Recovery time Disruption severity or performance loss Resilience triangle Considering that company performance is measured at the end of each period t (between t 0 and t 1 ), a curve is generated with the performance along time ( P it ). If there are no risks, the performance level of each company i is given by P i When a company is affected by the risk, a triangle pattern emerged showing the loss of company performance. However, some periods after the company performance recovers to the initial state P i

Page 30 Resilience Index To compute the resiliency index, we need to calculate the triangle area: 𝑅 ௜ ൌ 1 െ ׬ 𝑃 ௜ െ 𝑃 ௜௧ 𝜕𝑡 ௧ భ ௧ బ 𝑃 ௜ ሺ𝑡 ଵ െ 𝑡 ଴ ሻ where 𝑅 ௜ : the resilience index of organisation (or road, infrastructure,…) i , a value between 0 and 1. The resilience index of 0 means that there is no resilience to the disruption. 𝑃 ௜ : the performance level when it is not affected by the negative effects of a risk 𝑃 ௜௧ : the performance level in time period t 𝑡 ଴ : the lower limit of the time period based on which the company resilience index is determined; usually prior to the time instant at which the performance level is affected by the negative effects of the risk 𝑡 ଵ : the upper limit of the time period based on which the company resilience index is determined; generally corresponds to a time instant at which the performance level is already recovered from the negative effects of the risk

Page 31 Example A supply chain experiences a disruption on June 23, 2020, caused by a supplier disruption due to the COVID-19 pandemic which is the day #175 in the year 2020. Due to the disruption, a shortage of materials of 140 units is observed. The disruption is expected to be recovered in 130 days. During the disruption period, the material shortage should be compensated by means of redirecting the material flows to a backup supplier. The firm considers two alternative backup suppliers, A and B. Supplier A is capable of delivering 20 units within next 25 days after the disruption, 30 units within next 15 days, 40 units within next 30 days, and 50 within next 40 days. Supplier B can deliver 70 units within next 10 days after the disruption and 70 units within next 20 days. • Compute resilience for both supplier A and supplier B cases. • Which supply chain recovery strategy would entail a high resilience: supplier A or supplier B?

Page 32 Solution If recovery by supplier A is selected: 𝑅 ௜ ൌ 1 െ 20 ൈ 25 ൅ 30 ൈ 15 ൅ 40 ൈ 30 ൅ 50 ൈ 40 140 ൈ 130 ൌ 1 െ 10,000 18,200 ൌ 0.45 If recovery by supplier B is selected: 𝑅 ௜ ൌ 1 െ 70 ൈ 10 ൅ 70 ൈ 20 140 ൈ 130 ൌ 1 െ 2,800 18,200 ൌ 0.85 We can observe that the recovery strategy with redirecting material flow to supplier B implies a higher supply chain resilience. 𝑅 ௜ ൌ 1 െ ׬ 𝑃 ௜ െ 𝑃 ௜௧ 𝜕𝑡 ௧ భ ௧ బ 𝑃 ௜ ሺ𝑡 ଵ െ 𝑡 ଴ ሻ

Page 33 Quantifying resilience as a measure of service level (useful for network analysis) Suppose there are n nodes (suppliers) in the supply network, and the resilience index of node n denoted by R i is measured by: 𝑅 ௜ ൌ 1 െ ׬ 𝑆𝐿 ௜ െ 𝑆𝐿 ௜௧ 𝜕𝑡 ௧ భ ௧ బ 𝑡 ଵ െ 𝑡 ଴ ൌ 1 െ ∑ ሺ 1 െ 𝑆𝐿 ௜௧ 𝑆𝐿 ௜ ሻ ௧ భ ௧ୀ௧ బ 𝑡 ଵ െ 𝑡 ଴ 𝑆𝐿 ௜ : the service level when it is not affected by the negative effects of a risk 𝑆𝐿 ௜௧ : the service level in time period t Fraction of loss So, both metrics are bounded between 0 and 1. Higher values of R i means higher resilience .

Page 34 A bi-criteria method to measure supply chain resiliency We use attainable sets from control theory to measure supply chain resilience as a reaction to variations in both supply chain design and recovery control policies. The resilience loss is then calculated as the area of intersection of two rectangles , i.e., an approximated attainable set and the extremal limits of two performance indicators, e.g., service level and profit. The first (gray) rectangle is constructed on the basis of the extremal (e.g., minimum or maximum) values of the performance indicators (e.g., the minimum service level for a supply chain). After running the control algorithm with different recovery strategies for different execution scenarios, different attainable sets will be obtained (i.e., the sets that include all possible performance outcomes for different scenarios – the blank rectangles).

Page 35 The greater the intersection region of the two rectangles, the less resilient is the supply chain. Ideally, the square of intersections of the two rectangles should be zero, meaning that a supply chain is capable of withstanding to all disruption scenarios considered. The larger the distance between the two rectangles, the more unnecessary (i.e., excessive) redundancy the supply chain contains. Non-resilient case Resilient case

Page 36 Example Target performance of a supply chain is sales of 30,000 units and profit of 8 million AUD. Two supply chain network designs with associated backup recovery strategies are evaluated for resilience. Simulations of four different disruption scenarios for each of the two network designs resulted in the following performance outcomes: Network design I: Scenario 1: profit = 11,000; sales = 42,000 Scenario 2: profit = 10,000; sales = 40,000 Scenario 3: profit = 7000; sales = 38,000 Scenario 4: profit = 5000; sales = 27,000 Network design II: Scenario 1: profit = 12,000; sales = 37,000 Scenario 2: profit = 9000; sales = 35,000 Scenario 3: profit = 7000; sales = 29,000 Scenario 4: profit = 5000; sales = 18,000

Page 37 Example Target performance of a supply chain is sales of 30,000 units and profit of 8 million AUD . Network design I: Scenario 1: profit = 11,000; sales = 42,000 Scenario 2: profit = 10,000; sales = 40,000 Scenario 3: profit = 7000; sales = 38,000 Scenario 4: profit = 5000; sales = 27,000 Network design II: Scenario 1: profit = 12,000; sales = 37,000 Scenario 2: profit = 9000; sales = 35,000 Scenario 3: profit = 7000; sales = 29,000 Scenario 4: profit = 5000; sales = 18,000 8 6 4 2 0 0 5 10 15 20 25 30 Profit Sales Example for network design I & scenario 4 Intersection square Area = Resilience loss Area = (30-27)×(8-5)=9 Area = 0 Area = 0 Area = 0 Area = 0 Area = 0 Area = (30-29)×(8-7)=1 Area = (30-18)×(8-5)=36 The higher the value of the intersection area, the lower the resilience

Page 38 A few points: In reality, there are multiple outcomes of possible attainable performance in each scenario subject to variations of uncertainty factors and recovery policies. These sets of the outcomes form attainable sets that are then approximated (e.g., in form of rectangles). The selection of a particular point on the approximated attainable set (i.e., best case, worst case, average) depends on the risk aversion of a decision-maker.

Page 39 Recap: how to use the resilience estimates? • The resilience metrics can be used to compare resilience of different supply chain designs or the effectiveness of some recovery strategies. • The metrics are of value when comparing the firm’s resilience over several years. • Based on the resilience metrics, one can judge whether a resilience of a particular supply chain is acceptable or non-acceptable. • In order to identify the “acceptable” values or ranges of resilience (say 0.95 or between 0.93 and 0.98), the latter analysis is usually supported by an associated performance analysis, e.g., demand fulfillment or service level. • If a supply chain is capable of achieving some desired service level under some disruption scenarios and recovery actions, the resilience values can be considered satisfactory.

Page 40 Supply chain structures Echelon 1 Echelon 2 Echelon 3 Echelon 1 Echelon 2 Echelon 3 Echelon 4

Page 41 Ripple Effect in Supply Chain Ripple effect describes the impact of a disruption propagation on supply chain performance and disruption-based scope of changes in supply chain structural design and planning parameters. The Ripple Effect Exposure (REE) quantifies the Ripple Effect by combining features such as impacts of financial, customer, and operational performance, consideration of multi-echelon inventory, disruption duration and supplier importance. REE extends the analysis toward a multi-echelon setting, taking into account individual supplier exposure assessments.

Page 42 Supplier exposures to risk are computed at each supplier. Assessment of supplier exposure to risk is based on the analysis of “possible maximum loss” (PML), resulting from upstream disruptions in the supply chain. 𝑅𝐸𝐸 ൌ ෍ ෍ 𝑞 ௟௝ 𝑄 ௟ ൈ 𝑉 ௝ ൈ 𝜋 ௝ ൈ 𝑝 ௝ ൈ max ሺ 𝑐 െ ෍ 𝐼 ௞ ௦ ೕ ௞ୀଵ ; 0 ሻ ௠ ௝ୀଵ ௦ ௞ୀଵ j : product index, j ∈ [1,…, m ] 𝑙 ௝ : a part needed for product 𝑙 ௝ ∈ [1,…, 𝑟 ௝ ] 𝑞 ௟௝ : the number of units of part 𝑙 ௝ sourced from a supplier at stage k or the number of missing parts resulting from supplier disruption when considering stage k +1 according to downstream disruption propagation in the supply chain the total number of units of part 𝑄 ௟ : the total number of units of part sourced across all suppliers 𝑉 ௝ : the demand for product j 𝜋 ௝ : the profit margin for product j in % of revenue 𝑝 ௝ : the sales price of product j 𝑐 : the business interruption time 𝐼 ௞ : inventory (measured, e.g., as weeks of supply) held at the supply chain echelon 𝑘 : the supply chain echelon, k ∈ [1,…, j ]

Page 43 SEs are computed at each supplier. REE extends the analysis toward a multi- echelon setting. 𝑅𝐸𝐸 ൌ ෍ ෍ 𝑞 ௟௝ 𝑄 ௟ ൈ 𝑉 ௝ ൈ 𝜋 ௝ ൈ 𝑝 ௝ ൈ max ሺ 𝑐 െ ෍ 𝐼 ௞ ௦ ೕ ௞ୀଵ ; 0 ሻ ௠ ௝ୀଵ ௦ ௞ୀଵ Business financial impact value supplier importance ratio maximum time the system is affected lost sales due to the supplier disruptions business interruption time business continuation time, or the amount of time the focal company can continue to meet demand in spite of a disruption in supply REE is considered as a compounding function for the disruption propagation (i.e., the ripple effect) downstream in the supply chain, taking into account individual supplier exposure to risks. Depending on the decision-maker’s risk aversion, REE can reflect either the total impact, as a worst-case scenario where all suppliers would experience a disruption, or the average impact among all possibly disrupted suppliers.

Page 44 Example A three-stage retail supply chain with five suppliers, one distribution center (DC), and two customers with equal demand is considered. Each supplier delivers exactly one product (i.e., water, drinks, juice, yogurt, and milk) to the DC, and the DC delivers these products to customers. Analyze the REE with consideration of total impact and the importance of each homogeneous node in the supply network. Milk Yogurts Juices Drinks Water Parameters 850 1,200 980 1,900 1,310 Avg demand per day 30 40 30 50 60 Avg profit per unit 440 386 442 480 282 Number of units procured from supplier per day 850 1,200 980 1,900 1,310 Total number of units for parts across all suppliers 1,700 2,400 1,960 3,800 2,620 Total number of units for parts across all DCs 20 20 30 30 30 Avg inventory of suppliers in days 5 5 5 5 5 Avg inventory of DC in days 40 15 50 40 10 Max disruption duration due to COVID lockdowns

Page 45 Solution Milk Yogurts Juices Drinks Water Parameters 850 1,200 980 1,900 1,310 Avg demand per day ( 𝑉 ௝ ) 30 40 30 50 60 Avg profit per unit ( 𝜋 ௝ ) 440 386 442 480 282 Number of units procured from supplier per day ( 𝑞 ௟௝ ) 850 1,200 980 1,900 1,310 Total number of units for parts across all suppliers ( 𝑄 ௟ ) 1,700 2,400 1,960 3,800 2,620 Total number of units for parts across all DCs ( 𝑄 ௟ ) 20 20 30 30 30 Avg inventory of suppliers in days ( 𝐼 ௞ ) 5 5 5 5 5 Avg inventory of DC in days ( 𝐼 ௞ ) 40 15 50 40 10 Max disruption duration ( 𝑐 ) Business financial impact value Echelon Milk Yogurts Juices Drinks Water 13,200 15,440 13,260 24,000 15,720 Level supplier 6,600 7,720 6,630 12,000 7,860 Level DC maximum time the system is affected Milk Yogurts Juices Drinks Water 20 - 20 10 - 15 - 15 5 - Supplier exposure 769,200 258,450 The REE of the supply chain is $1,027,650, which is the maximum possible financial loss that our supply chain can suffer from supplier disruptions. So, we quantified the costs of disruptions.

Page 46 Supply chain Resilience Source: Ruel, S., El Baz, J., Ivanov, D., & Das, A. (2021). Supply chain viability: Conceptualization, measurement, and nomological validation. Annals of Operations Research , pp.1-30. Being lean, responsive, and globalized in structural designs, supply chains have also learned a great deal about how to act in line with nature and societal interests (i.e., becoming sustainable), how to strengthen their resilience during disruptions triggered by severe natural or man-made disasters, how to recover and manage the ripple effects, and how to utilize the technological advancements or so-called Industry 4.0.

Page 47 Logistics Network Design Decision Supply chain design decisions focus on selecting the number and location of plants, warehouses, and other supply chain nodes. Typical management questions include: • How many and where should the DCs or warehouses be located? • Which product should be packed or unpacked in the DCs or consolidation centres rather than the shipper’s location? • What customers should be serviced from each DC/ • How many echelons the network does have? • What service providers and value-added services should be used to meet market requirements?

Page 48 Outline  Freight transport modelling  Supply chain resilience  Green logistics

Page 49 Analytical framework for green logistics A model has been devised to map the complex relationship between logistical activity and its related environmental effects and costs. These effects and costs mainly arise from freight transport operations and, for this reason, most of the boxes and links in the diagram are associated with the movement of goods. Source: Green Logistics : Improving the Environmental Sustainability of Logistics. Alan McKinnon, Michael Browne, Anthony Whiteing, and Maja Piecyk

Page 50 Policy measure fore green logistics Governments have a range of policy instruments that they can deploy to reduce the environmental impacts of freight transport and logistics. These can be divided into six broad categories: 1. Taxation: comprising fuel taxes, vehicle excise duty and road-user charges. 2. Financial incentives: could be capital investments, subsidising the use of greener freight modes or urban consolidation depots 3. Regulation: factors related to vehicle design, operation, the status of freight operators, the tariffs they charge and even the capacity of the freight sector 4. Liberalisation and privatization: by enabling freight operators and freight markets to compete more effectively 5. Infrastructure and land-use planning 6. Advice and exhortation: identifying and promoting best environmental practice in freight transport, often closely working with freight associations.

Page 51 Spatial structure of the supply chain Freight modal Vehicle routing Vehicle utilisation Duty reduction for alternative fuels Exposure to congestion Fuel efficiency CO 2 intensity of energy source Regulations relating to vehicle emissions Relaxation of night delivery restrictions Promotion of road telematics Relaxation of vehicle size/weight regulations Support for improved vehicle design Road pricing Transport modal transfer Revenue-support for alternative modes Infrastructure investment in alternative modes land-use planning controls key freight transport parameters government transport policy measures

Page 52 Spatial structure of the supply chain Freight modal share Vehicle routing Vehicle utilisation Duty reduction for alternative fuels Exposure to congestion Fuel efficiency CO 2 intensity of energy source Regulations relating to vehicle emissions Relaxation of night delivery restrictions Promotion of road telematics Relaxation of vehicle size/weight regulations Support for improved vehicle design Road pricing multimodal infrastructure Revenue-support for alternative modes Infrastructure investment in alternative modes land-use planning controls key freight transport parameters government transport policy measures Taxation policy

Page 53 Course Recap Week 1: You learned the logistics process, key actors, decisions and a list of most common KPI measures. • Workshop: Using Excel, you learned how to extract KPI measures from container data. Week 2: You learnt about the digitalisation trends of logistics, including IoT and Blockchain • Workshop: By playing a simple game, you learned how blockchain works. Week 3: You started to learn how to mathematically formulate real- world optimisation problems in logistics. You implemented a few heuristic approaches to solve “Transportation Problem” • Workshop: Using Excel solver, you learned how to solve Transportation Problem in Excel. Week 1 will not be assessed in the final exam as it was already assessed in Assignment 1. Week 2 will not be assessed in the final exam.

Page 54 Course Recap Week 4: You learned about the path-finding algorithms including shortest path problem, max flow problem, min-cost flow problem. • You learned how to use Dijkstra algorithm to find the shortest path. • Workshop: You learned the basic of AMPL, and solved an optimisation problem using AMPL. You now can use AMPL to solve a more complex shortest path problem. Week 5: You learned about Facility Location Problem for finding the optimum location of a warehouse and Knapsack Problem for optimum bundling of shipments. • Workshop: Using AMPL, you learned how to solve a capacitated and uncapacitated Facility Location Problem.

Page 55 Course Recap Week 7: You learned about Travel Salesman Problem and Vehicle Routing Problem. • Workshop: You practiced how to code VRP in AMPL and how to eliminate subtours. Now, you can use AMPL to solve a real-world routing and delivery problem with multiple vehicles and different capacities. Week 8: You learned about finding the optimum shipment size. You also learned how to forecast future demands of a product/price/… • Workshop: Using Excel, you now can forecast the future trend of sth given the seasonality variations. Week 9: You learned about the real-world logistics problems of Port Botany, e.g. how to estimate the rail capacity, how the operators work,… Week 9 will not be assessed in the final exam.

Page 56 Course Recap Week 10: You now have the big picture of all levels and types of freight and logistics models. You can dive deep into each of those models once you have a real project. Now, you know how to measure resiliency of a system and you can estimate the resilience of a logistics system or a transport network. You also learned about the key components of green logistics. • Workshop: You practiced how to code Min Cost Flow Problem in AMPL and how to make decision on upgrading various links of a network. This part will not be assessed in the final exam.