Abstract
This paper aims to investigate the mechanisms underlying freight disruptions and to capture the full complexity of a postdisaster freight system. Qualitative data were collected through 20 semistructured interviews with key informants in Aotearoa New Zealand’s fresh fruit and vegetables sector. The interview data were used to conduct a root cause analysis and build a Current Reality Tree (a tool prescribed by the theory of constraints). This diagnostic map captures the multifaceted reality of postdisaster transportation disruptions and the intricate sequence of interconnected events and constraints having an impact on the performance of a freight system. Thirty-two constraints affecting the functionality of freight transportation, nine root causes, four ultimate impacts on freight performance, and 42 cause-and-effect relationships between these various system components were identified. These findings show that building freight resilience requires a holistic and concerted approach that includes policy development as well as business and operational adjustments. This paper also expands the theory of constraints beyond its existing boundaries by applying it to freight transportation. In contrast to the existing literature that commonly uses mathematical modeling to study transportation complexity, this research presents a qualitative and rich depiction of the reality and innerworkings of freight disruptions. The results are immediately relevant and provide concrete explanatory and predictive insights for practitioners and policy makers.
Keywords
Freight transportation creates place utility (goods available where they are needed by people and businesses) and time utility (goods available when they are needed for consumption or further processing). To these ends, freight is moved with high efficiency through complex transportation systems made of nodes (e.g., ports and intermodal terminals), linkages (e.g., roads and railways), modes, and organizations ( 1 ). However, constraints affecting the performance of transportation operations are common and range from regular challenges dealt with in the normal course of business (e.g., road congestion) to more severe events triggering a series of cascading effects across the globe. The COVID-19 pandemic has brought the complexity of supply chain disruptions to everyone’s attention and highlighted the need to rethink freight systems ( 2 ). Consequently, freight resilience has become a prominent topic in business circles, policy development, and academia.
Managing freight constraints is particularly challenging in the wake of a disaster that creates severe disruptions in established transportation networks ( 3 ) and requires freight plans to be rapidly adjusted to maintain the continuity of operations. In such circumstances, it is common for transportation systems to face acute capacity shortages (e.g., insufficient road, carrying capacity, or both), a lack of visibility about incoming freight volumes, as well as delayed and inaccurate information about the accessibility of the transportation network ( 4 ).
In the academic literature, mathematical modeling is commonly used to investigate postdisaster freight operations, including multiagent simulation, complex network theory simulation, and system dynamics modeling. However, authors in these fields agree that models accurate and relevant enough to be of immediate practical use are difficult to construct owing to data intensity, model complexity, and computational effort. To overcome these challenges, models are often simplified, but in the simplification, immediate relevance is lost.
Taking a different approach, this paper investigates the value of qualitative research and of the theory of constraints (TOC) to capture the complexity inherent in freight operations in the aftermath of a disaster. TOC is a management theory widely used in practice to systematically analyze and optimize operational systems. Building on systems thinking, TOC looks at problems comprehensively and establishes the complex and multifaceted interconnections between the parts of the whole system ( 5 ). In other words, TOC opens a system’s black box ( 6 ) by uncovering its underlying mechanisms and supporting a fine-grained understanding of its multiple components and how they interact with each other. To do so, TOC provides a set of problem-solving and system-improvement tools collectively called the Thinking Process (TP) ( 7 ). The TP rigorously analyzes verbalized system problems ( 8 ) and is, therefore, consistent with a qualitative research approach.
Moving from its primary focus on manufacturing operations, TOC’s principles and tools have been applied in a variety of fields and functional areas, including supply chain management ( 9 , 10 ), strategy ( 6 ), medical services ( 11 ), and tourism ( 12 ). Although TOC’s TP appears to be adapted to investigate the complexity of a freight system, it has not yet been applied to specifically study the transportation of goods ( 13 ) and has not yet been used to investigate freight disruptions in the wake of a disaster. Therefore, this paper aims to
Establish the value of TOC’s TP to investigate the multiple constraints and mechanisms underlying freight disruptions in the aftermath of a disaster; and
Capture and visualize the full complexity of the constraints inherent in a postdisaster freight system.
To achieve these objectives, interview data (collected through 20 semistructured interviews) and qualitative logic-based modeling were used. The original contribution of this research is the Current Reality Tree, which presents a comprehensive and detailed understanding of the complexity of postdisaster freight operations. It captures the web of multifaceted, embedded, and interdependent system relationships underpinning freight disruptions and spanning operations, policy, and strategy. By representing causal relationships, the results are immediately relevant and provide concrete explanatory and predictive insights for practitioners and policy makers.
Complexity in Freight Systems: Current Literature
Complexity in a freight system results from extensive interdependencies between multiple interacting components. In particular, complexity emerges when freight actors, which have their own goals and make autonomous decisions, interact with and depend on each other ( 14 ). This complexity is compounded by multimodal infrastructure systems, uncertainty ( 15 ), and external factors like geopolitics or adverse weather. In complex transportation systems, problems are anchored in the structure of the system, making cause-and-effect relationships not immediately apparent ( 16 ). In the wake of a disaster, an already complex freight system is thrown into crisis and the recovery back to normal may be uneven, partial, and protracted ( 17 , 18 ). Postdisaster freight systems are complex systems in a state of flux.
Dynamic simulation-based models incorporating uncertainty are capable of replicating (with some simplification) the complexity of freight systems ( 19 ). Multiagent simulation ( 20 ), complex network theory simulation ( 21 ), and system dynamics modeling ( 22 , 23 ) are three types of dynamic simulation models that have been applied to freight transportation modeling. However, applications in the postdisaster freight domain are scant.
Multiagent Simulation
Multiagent models replicate the effects of actors (shippers, carriers, etc.) who continuously make and review decisions as they interact with each other, the infrastructure, and the environment. Each actor trades off costs and benefits incurred by the decisions made during a time period and then updates the decision-making logic for the next period. Crainic et al. discuss applications of multiagent simulation in freight transportation to evaluate a range of interventions (from operational to policy-related) at a city logistics and a national or international level ( 14 ).
In disaster management, multiagent simulation is popular for evacuation planning ( 24 ), but only selected works have focused on the postdisaster movement of goods. Fikar et al. ( 25 , 26 ) and Kaddoussi et al. ( 27 ) combine optimization models and multiagent simulation to address the distribution of relief items. However, the simulation presented in Fikar et al. focuses on the behavior of residents accessing relief and not on the behavior of freight transportation actors ( 26 ). Similarly, the agents in Kaddoussi et al. represent the decision makers in the crisis management supply chain and not freight transportation actors per se ( 27 ). The decision support tool in Fikar et al. models locations, requests, vehicles, and closed roads as autonomous agents, but the authors state that more behavioral research is required for their tool to be applicable to real-world cases ( 25 ).
Despite its popularity, multiagent simulation in freight transportation modeling faces two significant challenges. Firstly, the behavior representation of the actors is oversimplified, often resulting from a lack of data that describes real-world behavior. Joubert and de Waal explain that existing approaches that infer agent behavior from small datasets without understanding the reasons behind the decisions lead to models that are unable to accurately predict changes in agent behavior when the system is disrupted ( 28 ). Secondly, the complexity of these systems requires large-scale models that push the boundaries of available computational architectures ( 14 , 26 , 27 ).
Complex Network Theory Simulation
Complex network theory captures the complexity of real-world systems by representing them as an interconnected web of nodes and links. Nodes represent entities (e.g., airports, road intersections, and organizations) and links connect the nodes (e.g., airports are connected by a scheduled flight, road intersections are connected by a road link, and organizations are connected by business transactions). Complex network techniques are popular when studying how networks break apart after a disruption ( 29 ) and proposing network recovery strategies ( 21 ). In their review, Pan et al. show the popularity of complex network theory for studying the vulnerability and resilience of freight systems ( 30 ), including ocean shipping ( 31 ), air transportation ( 32 ), road freight ( 33 ), and rail transportation ( 34 ).
However, Pan et al. identify four elements limiting the ability of complex network theory to represent the full complexity of postdisaster freight disruptions ( 30 ). Firstly, most studies in the transportation domain do not capture the multilayered reality of multimodal transportation systems, likely because large-scale models with significant detail are mathematically complex and present a high computational burden (which is the second limitation). In addition, contextual information, such as population density and geography, is not easily included in complex networks even though these factors have a decided impact. Finally, complex network theory is, like most mathematical methods, a data-driven approach and the challenge of sourcing and processing a diverse set of data can hinder its application.
In addition to these limitations, complex network analyses are predominantly descriptive. Although they elucidate the structure and dynamics that emerge within real-world systems, the explanatory and predictive power of these techniques is limited. Causal mechanisms are not easily identified in complex networks. By contrast, system dynamics modeling is an approach that identifies causal relationships at the outset.
System Dynamics Modeling
System dynamics is characterized by its qualitative causal loop approach that describes how elements in a model (for example vehicles and road infrastructure) interact and affect each other. If enough data are available to quantify the interactions, the second step is to develop a quantitative stock-flow model that tracks how the capacities (stocks) of different elements are built up or depleted as a result of interactions with other elements ( 23 ).
Abbas and Bell were the first to explore the applicability of system dynamics to transportation modeling ( 16 ). In the decade that followed this study, system dynamics was used mostly to investigate the adoption of alternative fuel vehicles, strategic policy interventions, and infrastructure development and maintenance ( 23 ). By 2017, freight systems were still mostly passed over or included very basically in holistic transportation models ( 35 ). Since then, the focus on decarbonization in freight transportation has spurred more system dynamics studies that investigate freight transportation more deeply ( 22 ).
Besiou et al. demonstrate the suitability of system dynamics to address the complexity of humanitarian operations ( 36 ). With regard to postdisaster freight transportation, Peng et al. explore the impact of inventory and information-gathering strategies on effective relief supply when roads are damaged after a seismic event ( 37 ), and Diedrichs et al. quantify the impact of communication on disaster relief operations ( 38 ). Both studies develop extensive simulations to test a range of experimental scenarios. While they yield useful insights for debate, neither provide concrete guidance for a real-world case. In fact, Diedrichs et al. point out that considerable work would be required to define the parameters of the model if it were to be applied in reality; they also admit that despite the sophistication of their model, it remains a crude representation of the reality ( 38 ). In a survey of system dynamics applications in humanitarian operations, Besiou and Van Wassenhove point out that rigorous qualitative analyses are often enough to produce novel research insights and practical impact ( 39 ). Building on this argument and the above-mentioned limitations of mathematical modeling, this paper takes a qualitative, TOC-based approach to capture the complexity of a postdisaster freight system.
TOC and the Thinking Process
TOC is a systems thinking methodology initially developed to manage factory bottlenecks and improve production flows. It was popularized in management circles from the 1980s with the publication of The Goal ( 40 ). Four decades and over 1,000 academic journal articles later ( 41 ), TOC has become a mature, robust, and versatile methodology used in research and everyday management practice ( 42 ).
TOC is based on the fundamental concept of “constraint”: a factor that limits the performance of a system relative to its goal. As explained by Rahman, each system has at least one constraint (a system with no constraint would have unlimited performance) ( 7 ). A constraint can be a physical resource creating a bottleneck owing to insufficient capacity to meet the demand placed on it (e.g., the limited production rate of a machine). A constraint can also be related to the market (e.g., market demand is lower than the available supply capacity), skills and employment, policy and regulations, the nature of products, and ingrained beliefs or assumptions ( 9 ).
TOC works on the ontological assumption that the problems under study are of systemic nature and that a web of embedded and interdependent relationships interconnect the constraints within the system. TOC also assumes that most constraints are symptoms originating from a limited number of root causes (or core constraints) ( 43 ). At the prescriptive level, TOC provides a process of thinking and several structured tools and guidelines to solve system problems in a logical and systematic manner ( 5 , 44 ). In particular, as stated, TOC provides a set of tools, TP, which are used to guide change management by capturing the complexity of systemic problems, understand the causal relationships between the components of the system, and decide how to improve its overall performance.
TP prescribes five tools that are the complementary steps of a system-improvement process. These tools have different purposes, can be used separately, and are often applied independently in the literature ( 45 ). The first TP tool, the Current Reality Tree (CRT), is a diagnostic map that depicts the innerworkings of the system under study. It captures the multiple constraints affecting the performance of the system, establishes the cause-and-effect relationships between these constraints, and identifies the root causes and their impact on performance ( 8 ). A CRT was used in this research as it provided the best fit to address the second objective (capturing and visualizing the full complexity inherent in a postdisaster freight system). Additional details about the application of this tool are provided in the next section. The four remaining tools are beyond the scope of this paper as they are used to plan solutions and their implementation ( 46 ).
In the field of transportation, the application of TOC is limited to references to its underlying principles and the use of tools for flow optimization. TOC was applied to investigate train drivers as the weakest link in the rail transportation process ( 47 ), identify terminal bottlenecks and improve railroad transportation performance ( 48 ), optimize the turnaround time of aircraft ( 49 ), and increase the return loading rate of transportation vehicles ( 50 ). In addition, Zivaljevic uses TOC to minimize traffic congestion in a motorway system by regulating traffic on the segments with lower capacity ( 13 ). Although these studies are in the field of transportation, none of them focuses specifically on freight and the TP tools are not used. To date, the TP tools have not been applied to investigate the complexity of freight transportation or to identify the many constraints affecting the postdisaster movement of goods. This paper addresses this gap by building a CRT using interview data. Before doing so, the next section presents the context of this research.
Materials and Methods
Research Context
The context of this research is the domestic movements of fresh produce (fresh fruit and vegetables) in the wake of a disaster in Aotearoa New Zealand (NZ). This context provides rich and meaningful data on this paper’s subject matter because NZ is prone to natural hazards that cause widespread transportation disruptions and because efficient transportation is critical in the fresh produce sector.
NZ Fresh Produce Sector
The NZ economy is dominated by primary industries, including fruit, vegetables, dairy, meat, wool, forestry, and seafood. Among those products, fresh produce contributes NZ$7 billion to the NZ economy ( 51 ). Fresh fruit and vegetables present specific transportation challenges because of their intrinsic characteristics, namely seasonality, perishability, and the resulting short shelf life. Figure 1 shows the actors involved in the fresh produce supply chain from the growers (e.g., orchards and farms) to the point of final distribution. This figure focuses on the NZ domestic market only, which is the focal point of this paper.
Fresh produce supply chain.
Over 4,200 NZ growers produce a variety of fresh produce, including kiwifruit, grapes, apples, avocados, onions, peas, potatoes, squash, and cherries ( 52 ). After being harvested, fresh produce is sorted and packaged at a pack house. It is subsequently kept in a cool store facility before being sold to fresh produce wholesalers, such as T&G Fresh, MG Fresh Produce Group, Primor Produce, Fresh Direct, Seeka, and Cater & Spencer Group. Given the level of integration in the NZ fresh produce industry, businesses are often involved in multiple supply chain stages. For example, several of the above fresh produce wholesalers are also growers, have their own pack houses and cool storage facilities, and operate a fleet of trucks for transportation ( 52 ). Fresh produce wholesalers sell the fruit and vegetables to grocery distributors. From there, fresh produce is distributed to retailers, including supermarkets (75%), other grocery stores (20%), and food service providers (5%) ( 53 ). The main grocery retail groups in NZ are Foodstuffs and Woolworths, both operating under multiple supermarket, grocery store, and food service banners ( 54 ).
In addition to the actors represented in Figure 1, transportation operators play a significant role in the fresh produce supply chain. Trucking is the dominant mode of transportation. Since strict temperature controls maintain the quality of fresh produce, road transportation is carried out by specialized companies operating a fleet of chilled trucks ( 52 ).
NZ Freight System
Domestic freight is largely dominated by road transportation in NZ, with some 93% of the total tonnage transported by trucks ( 55 ). The busiest road, State Highway 1, runs over 2,000 km from the North to the South of the country and is often referred to as the spine of the freight system. NZ’s secondary roads are not adapted to extensive freight transportation ( 4 ). Alternative modes for domestic transportation include rail, which accounts for about 6% of the total freight tonnage and, to a lesser extent, coastal shipping ( 55 ). Five ferries connect the North and the South Islands across the Cook Strait between Wellington and Picton. Since these ferries carry trucks, trailers, and wagons, they are a critical extension of State Highway 1 and of the railway network. Figure 2 shows State Highway 1 and the multiple locations mentioned throughout this paper.
Map of Aotearoa New Zealand (NZ) (created using the Eagle Technology Web Map software).
NZ is prone to natural hazards, including cyclones, floods, and earthquakes, which cause widespread transportation infrastructure damage. Because of the mountainous geography of the country and the lack of alternative roads and transportation modes, extensive freight disruptions are common in the wake of a disaster. For example, in the aftermath of the 2016 Kaikōura earthquake, State Highway 1 and the railway were immediately closed in the upper South Island, and only fully reopened more than a year after the event ( 4 ). More recently, Cyclone Gabrielle and the associated torrential rains that hit NZ’s North Island in February 2023 caused large slips and washouts that resulted in the closure of 10 state highways and restrictions at 19 additional road locations across the transportation network ( 56 ).
Interview Data
Since a CRT is built from the analysis of verbalized perceptions and opinions about system problems ( 8 , 46 ), qualitative data were collected through 20 semistructured interviews conducted with a range of informants (including growers, pack houses, cool stores, fresh produce wholesalers, transportation operators, grocery distributors, and retailers) from August to October 2022. The participants are representative of the main actors involved in the NZ fresh produce supply chain from the growers to the retailers, as presented earlier in Figure 1.
Semistructured interviews were selected as a data collection method because they enabled us to keep the conversation focused, while allowing for natural flow and the exploration of new topics. As discussed in the literature ( 57 , 58 ), semistructured interviews enable researchers to adjust their questions, the phrasing of the questions, or both, based on the context and the interviewees’ responses. This facilitates the understanding of each participant’s perspectives and experiences and makes it easier to delve into alternative but relevant areas of discussion. Although semistructured interviews lack standardization, which can lead to differences in the depth and breadth of the information collected, the level of flexibility offered was critical to collect rich data for this research.
A purposive sampling approach was taken. Accordingly, interviewees were selected based on their expertise and their ability to provide the most insightful information in relation to the research questions, rather than based on their job titles. In smaller businesses, the sales manager can also be the person in charge of distribution operations. Thirty-five percent of the interview participants were large businesses, 40% were medium businesses, and 25% were small businesses. In addition, interviewees were selected at the management level because managers have a broader view of freight transportation resilience and risk management. They can also provide insights into how decisions are made to overcome/mitigate freight disruptions in the wake of a disaster. Table 1 presents the interview participants and their positions.
Interview Participants
The interviews were conducted either in person, via videoconference, or by phone. One participant submitted answers to the interview questions in writing. As shown in Table 2, the questions were developed to prompt a variety of answers while providing focus to the discussions. The interviews were recorded and professionally transcribed to ensure an accurate record of the participants’ contributions. NVivo was used to code the transcripts.
Interview Questions
Note: NZ = Aotearoa New Zealand.
Root Cause Analysis
This section presents the step-by-step process followed to complete the root cause analysis as prescribed in the TOC literature and, ultimately, build the CRT that captures the complexity of postdisaster freight disruptions in the NZ fresh produce sector. The CRT is presented in Figure 3.
Step 1: Defining the System’s Boundaries, Purpose, and Performance
Before starting a system analysis using the TP tools, the TOC literature ( 8 , 59 ) recommends
Defining the boundaries of the system under study;
Defining the purpose of this system (what the system is designed to accomplish); and
Determining how this system assesses performance (what the system aims to achieve).
For this research, the system boundaries are defined as follows. The system under study is the NZ freight system with a particular focus on the domestic movements of fresh produce. The purpose of this system is to support the movement of goods from origin to destination within NZ. Performance is, therefore, reflected in the system’s functionality, namely its ability to create place utility and time utility. Time utility is particularly important for fresh produce deliveries because fruit and vegetables have a limited shelf life and are prone to spoilage if not delivered rapidly.
Step 2: Identifying the Issues Affecting the System’s Performance and Establishing Causal Relationships
As recommended in the TOC literature ( 8 ), our CRT was built by establishing an initial list of constraints, which was completed progressively as the chain of logic and the causal relationships between the elements of the system were explored. We worked upward from the natural hazard event to the ultimate impacts on performance. The typical TP graphical standards were used, including,
– Boxes representing the constraints;
– Arrows indicating causal relationships between the constraints;
– Ellipses representing the combined effect of constraints; and
– Square-cornered boxes representing the root causes.
Step 3: Reviewing and Revising for Clarity and Completeness
The cause-and-effect relationships represented in our CRT remained assumptions until they were systematically examined and challenged. To check their validity and make appropriate corrections, we used a set of logic rules called the Categories of Legitimate Reservation ( 5 ). According to these rules, assumptions were tested by identifying evidence substantiating the claims in the interview data. As prescribed by Scheinkopf, assumptions passed two tests before being validated: constraint existence (the existence of the constraints was questioned) and causality existence (each causal relationship between the constraints was questioned to ensure that the effect is the inevitable result of the cause) ( 8 ).
Adjustments were made as a result of this validation process ( 8 ), such as the inclusion of an additional cause if it was found that the cause was not the only reason for the effect, the combination of causes if the cause needed to exist in conjunction with another for the effect to occur, and the removal of a relationship if the cause did not result in the initially assumed effect.
Step 4: Identifying the Core Constraints
The final step of the process was to identify the core constraints (root causes) by looking for the entry points with no arrows pointing to them. TOC views these underlying issues as key leverage points for performance improvement ( 7 ). Since multiple problems experienced within a system are symptoms originating from a limited set of root causes, addressing these root causes can solve seemingly complex issues ( 60 ).
Results
This section presents the interview data used to build the CRT (Figure 3). To capture the interconnections between the elements of the system, the results are articulated around the following constraints and relationships:
Direct impacts of disaster events on transportation infrastructure;
Cascading effects on transportation operations and their ultimate impacts on freight performance (the chain of events affecting the performance of the freight system); and
Exacerbating factors (existing constraints in the NZ freight system and specific characteristics of fresh produce that exacerbate transportation disruptions).
To support the analysis, representative direct quotations are used. A reference reflecting the respondent category and the interview number (as shown earlier in Table 1) is added to each quotation. The respondent categories are abbreviated as follows: G for Growers, P for Pack Houses, C for Cool Stores, W for Fresh Produce Wholesalers, D for Grocery Distributors, R for Retailers, and T for Transportation Operators. These letters are combined when interviewees perform multiple activities. At the end of this section, Figure 3 presents the CRT built from the data by following the steps mentioned earlier.
Direct Impacts of Disaster Events on Transportation Infrastructure
The interview participants identified earthquakes as the natural hazard causing the most severe transportation disruptions over extended periods of time. This includes “the Canterbury and the Kaikōura earthquakes [that were] major events” (R-14). Interviewees also noted that transportation operations have been “affected more frequently in the last year than what we have experienced in the years before” (T-8). “Recently we have had quite a bit of flooding that has [had a] big impact on us nationwide” (T-17). Snow and storms were also mentioned as disruptive events (G/P/T/W-1).
Most interviewees emphasized the impacts of natural hazards on transportation infrastructure, particularly roads. For example, “during the Kaikōura earthquake, State Highway 1 to Picton was closed. So we went through the Lewis Pass. But when the earthquake first happened, we heard that that road was closed as well […] so it limits what you can do” (T-17). Like earthquakes, floods and storms result in roads being inaccessible: “we get quite a bit of flooding and especially heavy rain, which we’re expecting this week, and it could cause slips, close roads, and we can’t go anywhere” (T-5).
Weather events also affect the access to transportation services “between the North and South Islands, so across the Cook Strait. […] That can be a really rough piece of water and the ferries can’t sail” (W-10). Storms in the Cook Strait can trigger “unusually high waves over four meters” (T-16) and, in turn, “problems with the ferries, [so] we can't get [freight] over to the South Island” (G/P/T/W-1).
Cascading Effects on Transportation Operations and Ultimate Impacts on Freight Performance
Road is the dominant mode for the domestic transportation of fresh produce, with the Cook Strait ferries carrying the trucks between the North and the South Islands (T-5). Consequently, “anything that affects a main road, affects us” and disrupts the deliveries of fresh produce (P/C-12). In normal circumstances, road transportation enables same-day deliveries within the same island and overnight deliveries between the North and the South Islands (T-5, R-14). However, these transportation times increase in the wake of a disaster owing to the longer distances traveled on alternative roads. “If the Desert Road is closed through snow, [coming from Hamilton], we’ve got to go around by a different route, through New Plymouth. […] It doesn’t stop us, but it slows the delivery down” (G/P/T/W-1). Another interviewee explained that during the Kaikōura earthquake, “the only way was through the inland [roads] that added about three hours and a couple of hundred kilometers to the trip” (T-5). These increased transportation times result in “delayed deliveries to retail stores [and] getting products on shelves” (R-14). This affects transportation performance in relation to time utility, namely fresh fruit and vegetables are not delivered when they are expected.
Wastage because of the perishability of fresh produce can also happen. “Because it’s such short shelf life and quick turnaround, it needs to get there as quick as it can. […] If there’s a delay from a natural event, by the time it gets to our customer, it’s too old. […] It means we’ve got lots of wastage” (G-9). As summarized by an interviewee, “fresh produce doesn’t last long. The longer the roads are closed the more stock wastage we have” (G-18).
To avoid wastage and maintain the expected level of service despite the increased transportation times, additional vehicle and staff resources are needed.
For example, last year, there was a bridge south of Christchurch that was impacted and it added about two, three hours to the return freight run heading south. What that meant is drivers couldn’t get back […] within their driving hours. We had to put additional […] resources, so both fleet and people, to meet the demands of the network. […] Every additional kilometer we have to drive or hours we have to employ people, […] there is an increased operating cost (R-14).
Increased costs are also the result of the extra fuel needed when driving on longer alternative roads (G-2).
Because of the lack of redundancy in the NZ road network, alternative roads might not be available in the wake of a major disaster. “We don’t have a lot of alternate routes to travel. […] If one part of the network breaks there’s not a lot of options” (R-14). When critical infrastructure like key roads or the Cook Strait ferry terminals are unavailable, transportation operations are suspended: “recently, we had ferry delays to get across the Cook Strait. We had a backlog of trucks and […] products sitting in trucks. […] We ended up not shipping for a while until that backlog was cleared” (P/C-12). As a result of the suspended transportation operations, resources (including vehicles and labor) become idle, which ultimately increases transportation times: “flooding might bring down a slip in the road, which holds up trucks, [resulting in] somebody sitting there or gear sitting around inefficiently, not being used” (T-17).
Another consequence of the suspension of transportation operations is the cancellation of scheduled deliveries, leading to fresh produce being sold off to other markets, which ultimately has an impact on the performance of transportation operations in relation to place utility (fresh fruit and vegetables are not delivered where they are expected). “[The trucks] can’t go at all, and they say, I’m sorry, we cannot deliver where it’s got to go today, and the delay is such that they have to bring it back and we have to find an alternative market for that produce” (G-2).
Exacerbating Factors
Perishability of Fresh Produce
Given its perishable nature and the resulting risk of spoilage, fresh produce has strict transportation time and temperature requirements. “We’re dealing with a highly perishable product, particularly around berry fruit season” (W-10). The shelf life of these products ranges from less than a week to 12 days (P/C/W-7, G-9). Therefore, temperature-controlled transportation ranging from 0°C to 14°C is required for all fresh fruit and vegetables (W-10, R-14, P/C-12, T-17). The strict transportation time and temperature conditions explain the limited number of chilled transportation companies operating in NZ. “There aren’t a lot of transportation operators that specialize in produce. Knowing that produce is high risk, there’s probably only half a dozen” (W-10).
Overreliance on Road Transportation
Consistent with the dominance of roads for domestic freight in NZ, trucking is mostly used for the domestic movements of fresh produce. This is because of the lack of competitiveness of rail and coastal shipping that have longer transit times and cannot move freight from the point of origin to the point of destination without transshipment (which, in turn, increases transit times and transportation costs). “Trucks run at a 100 or 90 [kilometers an hour and] it’s point to point” (T-16). The rail “network runs at 80 kilometers an hour” (T-16). It “is too time-consuming [and] with multiple handling points” (G-18). As further explained by another interviewee, if I want to go and load through the port at Marsden Point, first off, I’ve got to use the road from Kerikeri, which is a huge fruit growing and produce region, to Marsden Point. That’s over an hour away, on road, first. Then, I get to the port, I now have to handle that produce. And it’s New Zealand labor […] so, suddenly, I’m adding handling costs. […] The ship then comes down to Tauranga. I then have to unload [and] to put it onto a truck. […] So, per unit, on average, this ends up being four or five times more expensive than just putting it on a truck” (P/C/W-7).
Focusing more on transit times, an interview participant explains that from Auckland to Christchurch, fresh produce can be delivered by trucks within 27 h. Using rail or coastal shipping would take 2 to 3 days (T-5). The multiple handling points from origin to destination also increase the risk of damage to fresh produce. “Every time you handle the product, there’s another opportunity for something to go wrong” (G-2).
Along the same lines, rail and coastal shipping are modes offering limited service frequency in NZ. As a result, freight waits before being actually dispatched, which leads to longer transit times: “if I wanted to put [fresh produce] on a train, for example, they’re not going to leave till tomorrow” (T-5). Ultimately, since “supermarkets […] want to have their product leave whenever it is today and arrive tomorrow” (T-5), “road is our only option” (G-18).
Poor Quality of the Road Infrastructure
The poor quality of the road infrastructure increases the risk of infrastructure failure in the event of a disaster. “Our roading infrastructure is fairly limited, fairly basic, […] at a lower standard than in countries where they have more significant investment in that infrastructure” (R-14). The effects are more severe in rural areas where “the road network doesn’t hold up in small weather events and fails continually […], especially in the smaller and more inaccessible regions such as Gisborne. […] Newly sealed roads should not fail, [but] they continually do” (G-18).
Delayed/Inaccurate Road Access Information
In the wake of a disaster, information is needed to successfully design alternative transportation plans: “communication is the key to solving these transport problems. […] So, if there’s a problem and roads are shut, then obviously we are discussing with the customers before we look at every alternative. […] You need to be able to communicate with not only your customers, but also with all the guys that are out on the road and tell them what’s going on and what’s happening” (T-5).
Comprehensive, timely, and accurate information supports well-founded decisions: “communication is key. A decision is better than no decision, but in hindsight, often that may be not the decision you should have made, but at the time, it’s the only information you’ve got” (P/C/W-7). As further explained by another interviewee, incomplete information leads to poor decision making: “you need to get as much information as you can before you decide to head off in another direction, for example, on a 10-hour detour. This has happened to us. We heard that the road was closed […], so we went off on this big detour and then two hours later, you get the information that the road is open” (T-5).
Although accurate information about the condition of roads and their level of accessibility is critical, it is not always available in a timely manner (T-17). When timely information from government is not available, transportation operators often find alternative, unofficial sources of information: “we would be getting that information real time from the social media feed, which is the quickest way, as opposed to the internet [government website], which often takes another half an hour to an hour to be updated” (T-5). As explained by another interview participant, the sharing of accurate and timely information does not require sophisticated information systems: “I don’t think we need a sophisticated digital or communication system. We have a thing called a mobile phone. […] That communication is instant and it’s between the harvest team and the truck driver in the orchard. You can’t get faster than that” (W-11).
Limited Carrying Capacity and Labor Shortage
The interview participants mentioned the limited level of carrying capacity and the shortage of skills in the NZ freight industry. Firstly, the Cook Strait crossing capacity is limited because the ferries are reaching the end of their working life and, therefore, require more extensive repairs and maintenance: “we are a bit exposed there even outside of disasters, just with service failures […] and ferries breaking down” (R-14). As confirmed by another participant: “One time we were going to ship to the South Island. During that time the ferry was sent to Italy for maintenance, at the time when we only had one ferry left. The transit time increased by several weeks” (D-3).
In addition to the ferries, trucking capacity is limited by the small number of companies specializing in chilled transportation and the high level of truck maintenance required as a result of the poor quality of the road infrastructure. “There’s only a couple of big companies that do chilled freight, and I feel like they’ve got a bit of a monopoly” (R-20). In addition, refrigerated trucks are made unavailable when maintenance is required: “we do a lot of maintenance on our vehicles because of the state of the roads” (T-17). The shortage of labor in the NZ transportation industry also presents a significant challenge: “with the lack of labor in NZ, […] we’ve had disruptions for the last three years” (P/C/W-7). In particular, “drivers are very hard to find, it’s hard to find people, especially on a short timeframe” (T-5).
Figure 3 presents the CRT built from the interview data following the step-by-step process explained earlier. Although the TP literature prescribes that all constraints should be written using complete sentences, this rule has not been applied in Figure 3 to simplify the presentation of the CRT. This CRT, which is further analyzed in the next section, provides an in-depth understanding of the sequence of interconnected events and constraints linking a major natural hazard occurring in NZ to poor transportation performance in the fresh produce sector.
Current Reality Tree.
Discussion
Using TOC’s TP to Investigate Postdisaster Freight Operations
This paper establishes the value of the first TP tool (CRT) to investigate the multiple constraints and mechanisms underlying freight disruptions in the aftermath of a disaster. It also captures and visualizes the complexity inherent in a postdisaster freight system. Figure 3 depicts 32 constraints affecting the functionality of the transportation system, four ultimate impacts on freight performance, and 42 cause-and effect relationships between these system components. The constraints identified are diverse and include physical constraints (e.g., road closures), capacity constraints (including the lack of redundancy in the road network and the limited carrying capacity available), market constraints (e.g., labor shortages), and operational constraints (e.g., the strict transportation time and temperature requirements for fresh produce). In doing so, this paper confirms that a CRT is useful to identify a range of physical and nonphysical constraints within a system ( 45 ).
Ultimately, a CRT offers the opportunity to look at freight operations as a whole system and from the perspective of what this system is designed to accomplish: moving goods and making them available when and where they are needed (place and time utility). It is valuable to capture and visualize in one diagram the chain of interconnected elements leading to poor transportation performance.
As shown in Figure 3, a severe natural hazard does not only affect place and time utility, but also leads to additional operational costs and fresh produce wastage. These results align with the broader literature on logistics performance that identifies the minimization of costs and the deliveries of goods meeting the expected quality requirements as key performance objectives ( 61 ). For over 30 years, research has argued that the measurement of logistics and transportation performance should be comprehensive enough to cover all aspects of efficiency and effectiveness ( 62 ). Most academic studies investigating logistics and transportation performance have taken this multidimensional approach, for example by developing and testing algorithms that combine both distribution costs and delivery reliability ( 63 ). When focusing more specifically on the performance of fresh food transportation, researchers have used models minimizing the various transportation costs while maintaining the quality of perishable products ( 64 ). Our research is in line with these studies. However, the interview participants did not discuss some of the additional performance aspects found in the literature, including social objectives ( 65 ) and environmental objectives through the measurement of fuel consumption and carbon emissions ( 66 ). Similarly, the interview data do not provide information on the broader economic impact of transportation and its contribution to economic competitiveness ( 67 ).
Mitigating the negative transportation performance impacts of a disaster requires identifying and tackling the root causes. Figure 3 shows nine root causes represented by the red-outlined, square-cornered entry points. As explained earlier, TOC states that a system’s overall performance is limited by a small set of root causes ( 8 ). Although it remains unclear what an acceptable number is, this research confirmed that TOC enabled the identification of a small number of core constraints. Our analysis generated a variety of root causes because the study included multiple stakeholders, in contrast with the existing literature that rarely goes beyond the boundaries of a single organization ( 60 ). This variety confirmed that building resilience in a freight network requires a holistic and concerted approach that includes policy development as well as business and operational adjustments ( 68 ).
The first root cause (“severe natural hazard occurring in NZ”) is beyond anyone’s control since earthquakes, storms, floods, and other events are caused by natural forces. However, the eight other root causes are opportunities for systemic improvement and can, to some extent, be addressed to increase the resilience of fresh produce transportation in NZ.
Although the second root cause (“perishability of fresh produce”) is an inherent characteristic of fruit and vegetables, precooling and packaging techniques can be used before transportation to extend shelf life. Precooling is the process of rapidly lowering the temperature of fresh produce immediately after harvest to slow down natural deterioration ( 69 ). Various food packaging techniques can also be used to preserve quality and marketability over time, including nanostructured coatings preventing the growth of micro-organisms leading to spoilage ( 70 ). Alternatively, antimicrobial edible coatings made from natural compounds can be applied to the surface of fresh produce to inhibit the growth of bacteria and fungi ( 71 ).
The third and fourth root causes (“the lack of alternative roads” and “the poor quality of the NZ road infrastructure”) show that the lack of redundancy in the NZ road network and the high level of damage repeatedly caused by natural events have severe impacts on freight movements and create vulnerabilities in and beyond the fresh produce sector. Government policy has a critical role to play here. This includes defining the critical infrastructure assets, identifying the vulnerabilities in the existing national freight system as well as the multiple layers of disruptions, and establishing priorities for transportation infrastructure improvements and developments to maintain the continuity and functionality of the transportation system in the wake of a major disaster. The CRT presented in Figure 3 clearly illustrates the complex relationships between public decision making (in particular in regard to transportation infrastructure development and maintenance) and the performance of private transportation operations ( 72 ).
Like roads, the Cook Strait ferries are critical infrastructure in NZ as they form a bridge between the North and the South Islands. For that reason, the fifth root cause (“Cook Strait ferries at the end of their working life”) needs to be addressed with plans to replace the existing ferries to provide reliable and resilient inter-island transportation.
The sixth and seventh root causes (“limited frequency of rail and coastal shipping services” and “multiple handling points in rail transportation and coastal shipping”) point to the importance of developing alternative transportation modes (including rail and coastal shipping for countries with long coastlines like NZ), integrating operations across these modes, and ensuring a high level of service frequency to reduce delivery times. Addressing these root causes calls for improved interfaces between transportation modes to increase interconnectivity and, ultimately, enable swift, frequent, seamless, and reliable transfers of goods between them. A unified and coordinated transportation system enables freight to move efficiently and without delays between modes and, ultimately, to mitigate the impact of disruptions ( 73 ).
The eighth root cause (“Delayed/inaccurate road access information”) has been highlighted as a key impediment to the efficient and effective reconfiguration of transportation operations in the wake of a disaster ( 4 , 74 ). Although the use of information technologies seems to be an obvious solution to increase visibility on freight movements, implementing these technologies is not without its challenges owing to the vast disparities in motivation and capabilities across businesses ( 75 , 76 ). This issue is all the more acute in the NZ context where the actors involved in the fresh produce sector are of very different sizes and have very different skills.
The ninth root cause (“labor shortage”) is not limited to the NZ freight industry. The practitioner and academic literature have discussed the challenges faced by businesses worldwide in finding and retaining workers, including truck drivers, owing to the aging of the current workforce, low salaries, and difficult working conditions ( 77 ). Strategies to address these issues include expanding support programs, educating and training freight workers, encouraging diversity, implementing flexible working conditions, and using autonomous vehicles ( 78 ).
CRTs and Mathematical Modeling
The disconnect between mathematical modeling and practice has been discussed earlier in the paper. The lack of practical relevance is often the result of complex simulation models beyond the grasp of practitioners with limited mathematical expertise, or oversimplified models resulting from the lack of data or computational power. Addressing these limitations, the CRT presented in Figure 3 was built without collecting and processing a diverse set of quantitative data or, as is usually required in simulations, estimating data when data do not exist. There was also no need to formulate, code, debug, or validate a large-scale mathematical model. In addition, access to significant computing power was not required.
Yet, our CRT presents in one diagram a comprehensive and multilayered depiction of the reality of postdisaster freight operations and, in the words of Boyd et al., opens the system’s black box ( 6 ). As per the classification of models according to the model use developed by Pidd ( 79 ), our CRT is a model that allows system investigation and improvement. This model is immediately graspable, that is, niche expertise is not needed to translate the output into user-friendly information. Therefore, this paper’s qualitative approach based on an everyday management tool integrated into academic research has achieved its goal: it captures the complexity of a postdisaster freight system and provides concrete and immediate insights for practitioners and policy makers, as further explained in the next section.
Contributions, Limitations, and Further Research
This paper expands TOC beyond its existing boundaries by applying it to freight transportation. In particular, it establishes the applicability and value of a CRT to systematically analyze the multiple constraints in a postdisaster freight system and the cause-and-effect relationships between them, and to identify the underlying systemic issues affecting transportation performance. Ultimately, the original contribution of this research is the CRT that provides in one diagram a comprehensive and fine-grained understanding of the complexity of postdisaster freight operations. By using interview data and qualitative logic-based modeling to capture this complexity, this paper presents a model that is not constrained by the mathematical representation of variables. Rather, the CRT presented in Figure 3 tells the story of why fresh produce is not available where and when expected in the aftermath of a disaster. It captures the web of multifaceted, embedded, and interdependent relationships interconnecting the elements of the system and spanning across operations, policy, and strategy. Mathematical modeling cannot replicate this complexity owing to the unavailability of the data needed to quantify the cause-and-effect relationships and the lack of computational power. Therefore, this research demonstrates the value of qualitative research and management tools to understand how a postdisaster freight system operates, and contributes a representation of freight transportation complexity that complements the freight modeling research available in the literature.
This paper also makes a practical contribution by providing explanatory and predictive insights and, therefore, helping transportation infrastructure owners and users better understand the impacts of disasters. It can feed into their decision making because understanding the innerworkings of freight disruptions in the aftermath of a disaster is a prerequisite for effective strategy, practice, and policy development. In particular, by mapping out the logical connections between the multiple constraints in a postdisaster freight system, our model can be used by managers in (and beyond) the fresh produce sector to develop new freight resilience initiatives and understand the impacts of these initiatives on performance (both in relation to operational costs and delivery quality and reliability). The CRT presented in Figure 3 can also be used as a communication tool. By conveying complex information in a clear and visual way, the CRT can facilitate discussions with both frontline workers (e.g., truck drivers) and staff involved in the planning, implementation, and evaluation of freight resilience programs. In policy development, our qualitative logic-based model can be used as an analytical input to the policy planning process and can provide guidance to formulate hypotheses for the impacts of transportation policy interventions designed to increase the resilience of the freight system.
However, although the CRT presented in Figure 3 is grounded in the reality of postdisaster freight operations and is based on robust methods to establish causal relationships, we accept that our model is not enough to fully support evidence-based policy interventions. Well-informed and transparent transportation policy decisions require the power of numbers and mathematical modeling that help policy makers measure the impact of their potential decisions and support the allocation of public resources where the greatest impact in relation to resilience can be achieved. Therefore, to increase the predictive power of our CRT, future research should develop algorithms that quantify (some of) the causal relationships illustrated in Figure 3.
Another limitation of this study relates to the focus of the research on a specific segment of the NZ economy. As a consequence, Figure 3 presents a construction of the reality of postdisaster freight disruptions based on the perceptions and opinions of NZ organizations and in relation to the transportation of fresh produce. Even if the topics discussed are extremely relevant beyond NZ and the fresh produce sector, further research should investigate the impact of postdisaster freight disruptions in other industrial and geographical contexts.
In addition, this study only uses the first TP tool. Since a CRT is a diagnostic tool, it enables the identification of nine root causes and, therefore, what underlying issues can be addressed. However, this research does not propose compelling solutions. The four remaining tools should be used to provide complementary insights and complete the change management process by guiding the development of effective strategies and the implementation of actions that will mitigate freight disruptions in the wake of a major disaster. Although the use of additional tools is not within the scope of this study, forthcoming research will investigate the value of two further TP tools, namely the Evaporating Cloud to identify and resolve conflicting requirements between core constraints, and the Future Reality Tree to eliminate the constraints identified in a CRT.
Conclusion
This paper investigates the mechanisms underlying postdisaster freight disruptions in the NZ fresh produce sector. It captures different layers of complexity and identifies the elements affecting the functionality of the domestic transportation system as well as the intricate sequence of interconnected events and constraints affecting the performance of freight operations. By using a qualitative approach rather than mathematical modeling to capture freight complexity, this paper provides an alternative and complementary representation of the cause-and-effect relationships between the elements of the system. The output of this research (the CRT presented in Figure 3) is within the grasp of practitioners with limited or no understanding of algorithms.
Footnotes
Acknowledgements
Author Contributions
The authors confirm contribution to the paper as follows: study conception and design: C. L’Hermitte, N. Trent; data collection: Y. Qin; analysis and interpretation of results: C. L’Hermitte, Y. Qin; draft manuscript preparation: C. L’Hermitte, N. Trent, D. Friday. All authors reviewed the results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by Te Hiranga Rū QuakeCoRE, a Centre of Research Excellence funded by the New Zealand Tertiary Education Commission. The publication number is 0870.