- Email: [email protected]
Location analysis for the modal shift of palletized building materials
Journal of Transport Geography 34 (2014) 44–53
Contents lists available at ScienceDirect
Journal of Transport Geography journal homepage: www.elsevier.com/locate/jtrangeo
Location analysis for the modal shift of palletized building materials Koen Mommens ⇑, Cathy Macharis Vrije Universiteit Brussel, MOBI, Pleinlaan 2, Brussels 1050, Belgium
a r t i c l e
i n f o
Keywords: Location analysis model Pallets Intermodal transport
a b s t r a c t At present, the distribution of palletized building materials is mostly carried out by trucks, despite their movements having negative effects on society, the economy and the environment. However, these problems can be reduced if the transport of palletized goods is shifted to inland waterways. By doing so, the goods are bundled for the main haulage by barge. In order to reduce the transport distances by truck to an absolute minimum, a possible last-mile distribution would have to be organized via a limited number of directly canal-served hubs. The locations of those hubs are crucial for the feasibility of modal shift. This study advances the transport geography literature by elaborating a location analysis model specifically for palletized goods. This model determines the optimal hub location by taking into account the large variation of origins and destinations of transport flows, while the introduction of a cost structure enables potential economic gains (cost savings) and reductions in CO2 emissions to be calculated. The analysis is performed for transport data on palletized building materials in Belgium. Two concepts were defined, which resulted in an optimal intermodal network of 9 hubs and one with 27 hubs; through the implementation of these networks, respectively 26% and 38% of the transport flows can be shifted to the inland waterways at a profitable cost. It can be expected that over time these percentages will increase further. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Most transport and mobility movements are performed on the road network (European Commission, 2011). Consequently, they are responsible for several negative externalities, such as congestion, pollution, noise nuisance and accidents. These externalities affect our daily lives, our economy and the environment. Other transport modes – like barge and rail – create similarly negative externalities, but in significantly lower proportions than road transport (CE Delft et al., 2011). As the demand for transport and mobility is expected to continue growing in the coming years and decennia, it is necessary to take into account those other transport modes as reasonable alternatives to the dominant road transport. For several Western European countries, inland waterways offer many opportunities, as its network is wide and mostly underutilised. In the past, large numbers of merchant transport operated within this network, with all kinds of goods transported by barge to the inner centre of Western European cities. In later ages, this practice disappeared, and canals were closed in many cities. However, following the containerisation of maritime transport that started in the 1970s, inland waterways were rediscovered in the 1990s by shippers, logistics service providers and ⇑ Corresponding author. E-mail addresses: [email protected] (K. Mommens), cathy.macharis@ vub.ac.be (C. Macharis). 0966-6923/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jtrangeo.2013.11.001
governments. After big bulk (as traditional first wave) and containers (as second wave) (VUB and COMiSOL, 2006), the challenge now is to find new types of loading units which can be transhipped to the waterway network. Palletized goods constitute one possibility in this regard. Road transport of palletized goods is currently carried out at a very competitive price, especially when the external costs are not included (Ricci and Black, 2005; Van Dorsser, 2004). However, some initiatives of transporting palletized goods via inland waterways have emerged in recent decades throughout Western Europe. During the early 2000s in the Netherlands, palletized drinks were transported via inland waterways. This Distrivaart project was abandoned in 2004, as the commercial basis was too small (Poppink, 2005). In another project in the Netherlands, the urban distribution of different kind of goods, including palletized goods, is organized by Mokum Mariteam via the local waterway network of Amsterdam. In the region of Paris (France), palletized building materials have been transported by barges since 1987 (Sétra, 2008). In Belgium also, the initial focus was oriented towards the construction sector, as a feasibility analysis (VUB and COMiSOL, 2006) indicated a clear potential for a modal shift of these goods. As a result of this study, different practical experiments were conducted throughout Flanders (Verbeke et al., 2007; VIM, 2012a,b). Currently, transport of palletized building materials on seven fixed routes connecting suppliers and customers that are directly adjacent to canals are being shifted to inland waterways.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
In order to organize the transport of palletized goods by barges in a larger and more (cost-) effective way, a network of transhipment hubs – or regional water-bound distribution centres (RWDCs) – needs to be created within the supply chain. Palletized goods can be bundled onto barges at those hubs, and more sustainable first- and last-mile distribution can be organized. In this way travel distances by trucks can be reduced to an absolute minimum. Previous theoretical studies (Cornillie and Macharis, 2006; Groothedde et al., 2005; Van Dorsser, 2004; VUB and COMiSOL, 2006) and practical experiments (Poppink, 2005; Verbeke et al., 2007, 2012; VIM, 2012a,b) have demonstrated that the length of both pre- and post-haulages has a large impact on the economic feasibility of the modal shift of palletized goods. The hub locations determine the lengths of both pre- and post-haulages: the closer the transhipment hubs are to the important volumes, the lower the number of pre- and post-haulage tonkm that needs to be performed by trucks. As a result, the economic, societal and ecological costs incurred from vehicular travel can be reduced. The need for a substantiated identification of the transhipment locations thus arises. Additionally, both private and public sectors are interested in the economic and ecologic potential of an implementation of the concept. In order to answer to the above-stated needs, we have developed LAMBTOP (Location Analysis Model for Barge Transport of Pallets). The model is constructed for the Belgian territory (Mommens and Macharis, 2012). This paper is structured as follows: in Section 2, we describe the challenges and principles of a modal shift of palletized goods via a network of RWDCs; the developed methodology is explained in Section 3. The obtained results based on data on palletized building materials transported within Belgium in 2011 are discussed in Section 4. We end with the conclusions. 2. Hub concept The integration of a main haulage by barge implies that suppliers and/or customers who are not located near an inland waterway need an initial and/or final haulage via road. In those cases, one can talk about intermodal transport. Still, the supply chain of palletized goods has very different logistical characteristics in comparison with inland waterway transport (IWT) in the form of bulk and containers. Table 1 illustrates these characteristics so as to indicate the challenges that must be dealt with when shifting palletized goods to the inland waterways. In order to match the characteristics of the palletized goods’ supply chains with those of the typical inland waterway transport, well-chosen transhipment hubs must be created within the supply chain. These hubs (or RWDCs) work as regional distribution centres whereby flows of palletized goods are bundled and transhipped to a barge. RWDCs can be supplied via road and inland waterways. Fig. 1 illustrates the possible flows of the supplier’s and/or customer’s warehouse, depending on the possible directly canal-served facilities:
Table 1 Supply chain characteristics of different cargo types. Source: Verbeke et al. (2012). Supply chain characteristic
Bulk
Containers
Pallets
Typical IWT
Number of SKU (stock keeping unit) Volume per SKU Speed of delivery
Few
No issue
Many
Few
High Low
No issue High
High Very low
Number of drops
Low
Low
Low Very high High
Very low
45
1. Directly canal-served supplier’s warehouse ) inland waterway ) RWDC ) post-haulage via road ) non-canal-served customer’s warehouse. 2. Directly canal-served supplier’s warehouse ) inland waterway ) directly canal-served customer’s warehouse. 3. Non-canal-served supplier’s warehouse ) pre-haulage via road ) RWDC ) inland waterway ) RWDC ) post-haulage via road ) non-canal-served customer’s warehouse. 4. Non-canal-served supplier’s warehouse ) pre-haulage via road ) RWDC ) inland waterway ) directly canal-served customer’s warehouse. 3. LAMBTOP methodology 3.1. Introduction Since in reality not every supplier or customer of palletized goods has a location near an inland waterway, a network of RWDCs needs to be created (VUB and COMiSOL, 2006). The optimal location of these RWDCs is crucial for all stakeholders, as these locations have a large impact on the profitability of the intermodal transports. Consequently, their locations will also define the potential turnover of the RWDC (Arnold et al., 2001; Aykin, 1995; Kayikci, 2010). The LAMBTOP (Location Analysis Model for Barge Transport Of Pallets) that will be discussed in this paper was developed to determine the optimal location of RWDCs in Belgium. Besides their locations, the model calculates the financial cost of the modal shift and the potential turnover for each distribution centre. This is important information given that transport costs are one of the main determinants in the intermodal decision-making process (Danielis et al., 2005; LOGIQ, 2000; Vannieuwenhuyse et al., 2003). Moreover, the transport price has proved itself as a clear bottleneck in previous studies, not to mention in the Distrivaart project (Poppink, 2005; VIM, 2012a; VUB and COMiSOL, 2006). In addition to the financial outcome, the model also calculates the potential reduction in CO2 emissions, so as to illustrate the ecological benefit of the modal shift. We focus on the financial cost structure in Section 3.2 before explaining the used methodology for CO2 comparison in Section 3.3. The methodology of the LAMBTOP is summarized in Section 3.4. Finally, we focus on the data collection in Section 3.5. 3.2. Cost comparison The calculation of the financial difference between unimodal road transport and intermodal transport is performed with a cost structure based both on theoretical analyses (Essenciál Supply Chain Architects, 2011; Freight Best Practices, 2005) and practical experiments with palletized building materials (De Munck, 2010; Verbeke et al., 2007; VIM, 2012b). Information regarding these experiments was obtained through contact with several transport experts1 that accompanied these field tests. In the first stage, the cost structure only considers direct transport-related costs. In subsequent stages, different scenarios have been added, allowing us to test the sensitivity of the analysis. Firstly, additional depot costs ranging from 1€/ton to 4€/ton have been introduced into the intermodal supply chain. A second scenario takes into account administrative savings accrued through saved transport documents, payment transactions and invoices. The last scenario tests the impact of introducing a general road
1 The transport experts assist Flemish companies in their search for possibilities to shift (portions of) their transport flows to the inland waterways. The advice and services provided to the transport companies by these transport experts are free of charge, as the experts are appointed by a cooperation of waterway administrators and enterprise unions.
46
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
Fig. 1. Supply chain of the transport of palletized goods by barge. Source: VUB and COMiSOL (2006).
pricing system conform the rate of the LWK-Maut in Germany (0.15€/km) (Blauwens et al., 2011). Management costs, investment costs and external costs (congestion, accidents, emissions, etc.) are not included. The financial costs are expressed in €/ton, whereas the distance dependent variables are expressed in €/tonkm. The time trade-offs created by the intermodal transport are not included. This is justified, as the construction sector’s interest and willingness in a modal shift of their flows prove that any gains in reliability and cost efficiency brought about by this shift are far more important than any created time trade-offs. The following cost structure is obtained for the supply chain of the unimodal road transport:
C R ¼ C LT þ C H dr þ C UT
ð1Þ
Under the assumption that both the supplier’s and the customer’s warehouses have no directly canal-served facilities, the cost structure for the intermodal transport’s supply chain, which combines the costs of the different steps, is defined as follows:
C I ¼ C LT þ C PH dpr þ C UT þ C LB þ C B db þ C UB þ C LT þ C PH dpo þ C UT
ð2Þ
The difference between the cost of the intermodal transport (CI) and the unimodal road transport (CR) is illustrated in Fig. 2, representing the basic structure of the cost comparison. Rate books reveal that transportation costs mainly depend on distance.
Fig. 2. Cost structure. Source: own composition.
Moreover, the relationship between those costs and distance is approximated adequately by using a linear increasing function (Daganzo, 2005). Both transports start with an initial cost of loading the truck (CLT). In the intermodal variant, the truck drives over a distance (dpr) to the optimal RWDC at a higher cost (CPH) than the cost of the main haulage by truck of the unimodal transport (CH). This is due to the assumption of a probable empty return haulage in the case of pre- and post-haulages (100% empty kilometres) (De Munck, 2010; Essenciál Supply Chain Architects, 2011), whereas the main haulages of the unimodal transport are assumed to be done at a ratio of 26.5% empty kilometres (Freight Best Practices, 2005). Once the truck arrives at the RWDC, it has to be unloaded (CUT), after which the palletized goods have to be loaded into a barge (CLB). The loading of the barge is assumed to be done by forklifts or mobile cranes. Both techniques have proved themselves as the most efficient during practical experiments where different loading techniques were tested (Verbeke et al., 2012; VIM, 2012b). Once the barge is filled up with palletized building materials to an average loading factor (weight) of 95%, it navigates to an average cost (CB) over a distance (db). Additionally, the assumption is made that the barge transport counts 24% empty kilometres. This percentage has been validated by a market player. Upon arrival at the second RWDC, the barge is unloaded (CUB) using the same techniques as when it was loaded. The palletized goods are then transhipped onto a truck (CLT), which brings them to the final destination – as we assume – at the same cost as the pre-haulage (CPH). The additional transhipment costs at the RWDC (CUT, CLB, CUB and CLT) have a large impact on the overall price of the intermodal transport. For the combination of one truck (un)loading (CUT) and one barge (un)loading (CLB), a break-even distance of 60.8 km is needed. The break-even distance is the distance at which the costs of intermodal transport equal the costs of unimodal road transport (Pekin et al., 2012; Rutten, 1998). In cases where both pre- and post-haulages are needed, the break-even distance for the transhipment costs made at the RWDC is 121.7 km. The costs of the pre- and post-haulages are not included in these break-even distances. One cannot calculate a general break-even distance, simply because it depends on the length of the pre- and/or post-haulages, which varies for every origin–destination combination.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
In a comparison of the main haulages, the financial cost of transportation by barge (CB) is more than half the financial cost of main haulage by truck (CH). The financial profit from an intermodal shift should thus be made in this section of the supply chain in order to compensate for the extra handlings at the RWDC and possible pre- and/or post-haulages (see Fig. 2). Although, as stated before, no general break-even distance can be calculated, several scenarios can be analysed for different scenarios with different pre- and/or post-haulages distances (see Fig. 3). The initial cost contains the costs of the pre- and post-haulages and all the transhipment costs of both barge and truck. As such, the break-even distance can be calculated for each scenario. In the case where both the supplier’s and customer’s warehouse have directly canal-served facilities – and consequently no preand post-haulage, and no additional transhipment costs of (un)loading the truck are needed – choosing the intermodal alternative is theoretically always a profitable one. If one pre- or post-haulage is needed, the extra initial costs of these haulages and loading and unloading of the truck must be taken into account. The break-even distance of the first scenario (with 5 km of pre- or post-haulage) is 68.1 km. In cases with a pre- or post-haulage of 30 km, the break-even distance rises to 140.4 km. The break-even distance of the minimum scenario of pre- and post-haulage (5 km) is 136.2 km. The maximum route distance via road between Belgian municipalities located within a buffer of 30 km of an inland waterway is approximately 250 km. Within this distance, all stated intermodal scenarios (up to 40 km of preand post-haulages) are profitable. More precisely, the tipping point for a break-even distance of 250 km is 46.8 km of pre- and posthaulage. It should be noted that in reality the distance of the main haulage by barge is often much longer than the main haulage by truck. This is largely due to the difference in density between the inland waterway network and road network. In the Belgian case, the difference in length of the main-haulage by barge and by unimodal truck is approximately 25% on average, which implies a decrease in the so-called tipping point to a more realistic 28.8 km. This distance corresponds to the 30 km mentioned in previous studies as the maximum distance of pre- and post-haulage for which the modal shift of palletized goods can be profitable (Cornillie and Macharis, 2006; Essenciál Supply Chain Architects, 2011; Poppink, 2005). 3.3. CO2 comparison In general, intermodal transport creates lower external costs (pollution, emissions, noise, congestion and accidents) than
47
unimodal road transport (CE Delft et al., 2011; Kreutzberger et al., 2006). In this paper, only the transport-related CO2 emissions are taken into account. Some assumptions have to be made in order to calculate the CO2 emissions. First of all, as the obtained data lacks information on vehicle type, the haulages via road are assumed to be done by a truck-trailer combination (36.1 ton average gross vehicle weight; 26.2 ton loading capacity). The CO2 emissions are based on European average figures (10% urban, 15% non-urban and 75% motorway). The study we used – the STREAM International Freight study (Den Boer et al., 2011) – focuses on different transport modes (rail, road, inland waterways and short sea shipping). The information on average truckloads enables us to calculate the CO2 emissions for all the haulages by truck (unimodal variant, and pre- and post-haulage). The pre- and post-haulages are assumed to have an empty return haulage (100% empty kilometres), while the main haulage in the unimodal variant is assumed to have 26.5% empty kilometres (Freight Best Practices, 2005). The CO2 emissions for barge transport are calculated on the bases of the Flemish Environmental Agency’s (Vlaamse Milieu Maatschappij (VMM)) calculations, using the EMMOSS model of Transport & Mobility Leuven (Van Lier and Macharis, 2013). Moreover, an assumption is made that transport by barge is carried out with an average load factor (weight) of 95%, and an average empty running of 24%. 3.4. Methodology The LAMBTOP is a geographic information system (GIS) based model, which consists of different network-layers, each representing a transport mode (road and barge). The locations of departures and destinations are connected to the network layers by their corresponding nodes. In many cases, the municipality centres – defined as the main church of the municipality – act as those nodes (Fig. 4). If it is known that a departure and/or destination location is directly canal-served, this is included as such in the analysis. The network for Belgium is built by combining the following digital databases: The inland waterways layer is extracted from the Environmental Systems Research Institute (ESRI) data set for Europe. The road layer and municipality layer are obtained from the MultiNet database of Tele Atlas. The model is set up for Belgium, but can easily be transformed for other regions, and preferably expanded to a European scale.
Fig. 3. Cost comparison of unimodal road transport versus intermodal transport (IT) scenarios. Source: own composition.
48
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
Fig. 4. The network in ArcGis. Source: own composition.
As a first step, the obtained geographical information about the flows of the palletized goods is uploaded to the model in the form of an origin–destination matrix (OD-matrix) (Fig. 5). These ODcombinations are linked to their corresponding nodes in the origin–destination-point layer. It enables us to identify and map the transport flows of palletized goods. The model identifies the OD-combinations and computes the unimodal routes travelled via the road network. These routes are calculated by a shortest time path algorithm; Dijkstra’s (1959) algorithm is used for every path algorithm in the entire analysis (unimodal routes, buffer, location analysis, intermodal routes). The second step is the distribution analysis of the transport of palletized goods, which will determine the future locations of the RWDCs. The distribution locations (the nodes of the departure and arrival locations) are weighted by the sum of the tonnage of
Fig. 5. The structure of the LAMBTOP. Source: own composition.
all the routes that start and arrive in these receptive nodes. Next, a precondition is formulated; the distribution locations used as the ‘market area’ for the location analysis of the RWDCs are limited to the nodes which are located within a predefined buffer of an inland waterway (illustrated in Fig. 6). The buffer is fixed once at 15 km and once at 30 km within this analysis, using the road network for a shortest time path algorithm. Although the critical maximum distance of the pre- and post-haulage necessary for the overall intermodal transport to be profitable depends on the overall transport distance and the used cost structure, it can be assumed that 30 km of pre- and post-haulage is the ultimate maximum for the Belgian case (Cornillie and Macharis, 2006; Essenciál Supply Chain Architects, 2011; Poppink, 2005; Section 3.1). By using this delimitation, intermodal routes with too long and consequently too expensive pre- and post-haulages will be excluded. The potential locations of the RWDCs are defined as locations on an inland waterway of minimal class III (P600 ton), and lying within a predefined distance (50 m) of a trafficable road. Thanks to this precondition, future RWDCs will not be located in a pedestrian city centre or in a protected nature reserve. Furthermore, RWDCs will already have a direct connection to the existing road network, thereby avoiding heavy investments in road infrastructure. The determination of the optimal locations (step 3) is based on the ‘Location–Allocation’ procedure of the ‘ArcGIS Network Analyst’ tool. The procedure starts with the calculation of the shortest path between every distribution location and every potential RWDC location using the road network and the algorithm of Dijkstra (1959). Then an edited version of the obtained cost matrix is constructed (Hillsman, 1984), which enables the heuristic to solve a variety of different problem types. Next, the location–allocation process generates a set of semi-randomized solutions, before applying the vertex substitution heuristic of Teitz and Bart (1968) to create a group of good solutions (Church and Sorensen, 1994). A metaheuristic then combines these good solutions to find better solutions until no additional improvement is found. Finally the metaheuristic delivers the best solution found (ESRI, 2010). The optimal locations of the RWDCs vary with the number of chosen RWDCs. This number is chosen on the basis of the ‘market share’ and their spatial distribution. In the concept which uses a buffer of 30 km (‘concept 30 km’), the idea is that it is better to enlarge one directly canal-served distribution centre than open two of them in the same ‘market area’. Thus the number of RWDCs will be limited. In the concept with a buffer of 15 km (‘concept 15 km’), RWDCs are seen as small transhipment platforms. This permits a larger number of RWDCs, which is interesting because it lowers the pre- and post-haulage distances and the captured volume. Therefore, the hubs may need to be combined with other logistic activities to achieve required volumes.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
49
Fig. 6. Distribution analysis (tonnage per municipality) and potential market area. Source: own composition.
Once the number and the optimal location of the future RWDCs is set, a GIS network is created specifically for the intermodal transport. It combines the road layer (for the pre- and post-haulage), the RWDC locations (as transhipment nodes) and the inland waterway network layer. Within this newly created intermodal network, a shortest time path algorithm for the pre- and post-haulages and a shortest route algorithm for the main haulage by barge are performed for every OD-combination. The respective distances of the unimodal routes by road and the intermodal routes are calculated for all the OD combinations, and the distances are then linked to the cost structure. A cost analysis is performed for every individual OD-route. The combination of all routes gives a global overview in which the routes where the modal shift is cost-efficient will describe a realistic potential modal shift and a realistic potential turnover (in tons) of the future RWDCs. Assuming that these profitable transports will be shifted towards the inland waterways, it is possible to calculate the saved truck movements, and consequently an estimation of the potential CO2 reduction. Different cost scenarios – an introduction of road pricing, an introduction of depot costs (1€/ton to 4€/ton) and administrative cost savings – enable us to test the sensitivity of the analysis. 3.5. Data collection The analysis is performed for Belgian transport flows of palletized building materials for the year 2011. The construction sector
is an import economic motor for the country, representing 7–8% of Belgium’s GNP. The sector also has a large impact on mobility, given that construction-related transport represents 25% of freight transport on Belgian highways (VIM, 2012b) and a large portion of this freight is loaded on pallets. In total, almost 53 million tons of palletized goods are transported within Belgian borders each year (ADSEI, 2010). It is impossible to identify the palletized building materials from these last data, but a feasibility analysis (VUB and COMiSOL, 2006) demonstrated that 6–7 million tons of this type of good could be shifted to inland waterways. The used data for 2011 were collected in the ‘Build Over Water’ project (VIM, 2012a,b). The goal of this project was to analyse the feasibility of a modal shift of palletized building materials to inland waterways. This was done through a combination of different practical experiments, and by a distribution and location analysis on the basis of the LAMBTOP. The flows were obtained via a dedicated survey in which 9 out of the 50 reported producers participated. Together, they represent 1.163 million tons of transported palletized goods; in other words, the survey covers approximately 1/6 of the total volume that could be shifted to inland waterways. The data contain information about the producer, the tonnage, and the origin and destination at the spatial scale of the municipality. For the directly canal-served sites, the address location is used as the geographic node. For all non-canal-served locations, or cases for which no address information is given, the tonnages were assigned to the municipality centre. In cases where a municipality contains both water-served and non-canal-served destinations,
50
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
Table 2 Overall potential turnover of RWDCs. Source: own composition. Concept 30
Concept 15
Location
Volume (ton)
Location
Volume (ton)
Location
Volume (ton)
Location
Volume (ton)
Kortrijk Antwerp Hasselt Aalst Mol Leuven Pont-à-Celles Bruges Liège
346,988 197,149 142,807 101,526 88,261 77,109 66,841 62,918 58,773
Wervik Waregem Dilsen-Stokkem Hasselt Beringen Willebroek Beerse Liège Temse
160,614 156,500 65,210 64,330 63,132 59,531 53,283 47,482 41,998
Riemst Leuven Herentals Antwerp Bocholt Halle Brussels Aalst Roeselare
37,131 36,600 36,446 29,106 29,013 28,081 27,452 25,711 25,710
Charleroi Jabbeke Gent Pecq Manage Aalter Lier Huy Saint Ghislain
23,951 23,398 23,305 18,381 17,273 11,331 10,213 3917 3766
the assigned tonnage is divided proportionally between these locations.
4. Results and discussion The 1.163 million tons of palletized building materials creates an origin–destination-matrix (OD-matrix) of 1462 combinations. This matrix was uploaded into the model and the distribution analysis is illustrated in Fig. 6. The production locations are – as origins – mainly concentrated in the west and north of the country. The tonnage of the destinations is more equally distributed, but some general larger concentrations can be distinguished in the Flemish
part of the country, the ABC axis (Antwerp–Brussels–Charleroi) and around the city of Liège. In order to limit the pre- and/or post-haulage distances – in prospect of the profitability of the modal shift – only the distribution locations which lie within the predefined buffer of 30 km (in the left map) and 15 km (in the right map) of an inland waterway are selected as the ‘market area’ for future optimal RWDCs. For the concept with a buffer of 30 km, the location analysis works out an optimal number of nine RWDCs. The buffers of 30 km around these nine optimal RWDC locations cover 96.5% of the 1.184 million tons which have been defined as the ‘market area’. A market area with less than nine RWDCs lacks the ability to serve considerable market areas within a reasonable pre- and post-haulage distance. Once
Fig. 7. Overall potential turnover of RWDCs and the intermodal routes. Source: own composition.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
51
Fig. 8. Volume and CO2 emissions savings for the profitable intermodal routes of each scenario. Source: own composition.
Fig. 9. Profitable intermodal routes for scenario with road pricing and depot cost (left: ‘concept 30 km’; right: ‘concept 15 km’). Source: own composition.
more than nine RWDCs are chosen, the added value of the new RWDCs becomes negligible. In the case of a market area of 15 km, the model returns 27 optimal RWDC locations. They cover 99.2% of the 1.122 million non-canal-served tons distributed within the market area. Four of the 27 optimal locations are closely located to a directly canal-served origin location. It can be expected that, in case of implementation, these neighbouring locations will merge in time to one transhipment location. This is an illustration that the optimal locations in practice will be adjusted to field characteristics, because currently – except for the proximity of a trafficable road – the location analysis does not take into account other land use and traffic factors. Public and private sector are aware of that, and consequently they are not only focussing on the exact
optimal locations, but if necessary also to suitable area’s near those optimal locations. Fig. 7 illustrates the spatial distribution of the optimal RDWC locations for both concepts (left: ‘concept 30 km’, right: ‘concept 15 km’). Additionally, the intermodal routes and the overall potential turnover (in tons) of the RWDCs are shown in Table 2. The overall potential turnover of a RWDC is the tonnage which would be transhipped there if all given OD-combinations are shifted to the inland waterway. Of course, this is not a realistic assumption because, according to the cost structure, the flows have to cope with a break-even distance (Section 3.2). The volumes per RWDC for the ‘concept 15 km’ are low, as they are seen as small transhipment platforms which may or may not be combined with other lo-
52
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
gistic activities to achieve required volumes. When applying the cost structure to the intermodal and unimodal routes, it can be seen in Fig. 8, that today over 300,000 tons of palletized building materials can be transhipped at a profitable cost via the 9 optimal RWDC locations to inland waterways. In the case of the 27 RWDCs (or ‘concept 15 km’), the profitable amount rises to almost 450,000 tons. These data are for the survey which covered approximately 1/6 of the total potential transport flows of palletized building materials. Therefore, it can be assumed that the real-life modal shift of palletized building materials will be larger than those stated tonnages. The potential savings in transport-related CO2 emissions are shown in Fig. 8. The total CO2 emissions savings for the profitable intermodal routes with respect to their unimodal alternatives varies between 232 and 922 tons, depending on the concept and scenario. The proportional ecological gains are clearly higher in the ‘concept 15 km’ scenario. This is also the case for the scenario where a general road pricing system that conforms to the rate of the LWK-Maut in Germany is implemented (Blauwens et al., 2011). For this scenario, the profitable modal shift increases by approximately 15%: to over 350,000 tons for ‘concept 30 km’ and almost 520,000 tons for ‘concept 15 km’. The introduction of administrative costs savings to the general cost structure has more or less the same impact as road pricing. When depot activity costs are added to the general cost structure, the profitability of the concept is reduced and added value needs to be created for depot activities. Added value can be created through additional services at the RWDC (VIM, 2012b) or through consolidation with other types of cargo, especially in the construction sector, where the industry is often located near the inland waterways and where bulk materials have a similar distribution pattern as the palletized cargo. The RWDC could thus also serve as a transhipment hub for bulk materials. Combining the transport of bulk and palletized cargo in one barge is, however, not suitable, as it enlarges the transhipment costs. Finally, the added depot activity costs can also be encountered by costs savings through consolidation of different deliveries into one post-haulage. The last scenario combines the road pricing scenario with a depot cost of 2€. Under these circumstances, 17.8% and 20.2% of the initial 1.163 million tons can be shifted at a profitable cost for ‘concept 30 km’ and ‘concept 15 km’ respectively. When mapping the intermodal routes of these profitable combinations (Fig. 9), it can be seen that they are mostly allocated to RWDCs in the west and the east of the country. Transport distances are therefore high, so the break-even distance is reached (Fig. 3). This fact advocates the introduction of an international RWDC network where average transport distances will be higher. Consequently, the financial and ecological benefits of the concept increases.
5. Conclusions The transport of palletized goods via Belgian inland waterways has evolved from a theoretical feasibility analysis (VUB and COMiSOL, 2006), which illustrated a clear potential for palletized building materials and fast moving consumer goods, to practical experiments and even different operational schemes for the transportation of palletized building materials. This multiplicity of experience has enabled the development in this paper of a cost structure based on transport-related financial costs. Different scenarios have been developed, including a general introduction of road pricing, administrative savings and depot costs. The cost structure shows an absolute economical potential for the modal shift of palletized goods transportation when both producer and customer are located near inland waterways. In cases where the supplier’s and/or customer’s warehouse is not located
near an inland waterway, RWDCs and pre- and/or post-haulages become inevitable. It is obvious that shorter haulage distances increase the (financial) feasibility of modal shift for palletized goods. So, the location choice of the RWDC, and consequently the minimization of pre- and/or post-haulages, is crucial for the success of the concept. In this paper, the LAMBTOP has been described. This model allows us to analyse the optimal RWDC locations. It has been developed to facilitate the modal shift via a network of hubs, and this for a new type of cargo (palletized goods) with specific characteristics. Moreover, the model takes into account the variation in origins and destinations to calculate these optimal locations. The model is also able to calculate the financial costs of the modal shift and the potential turnover of every RWDC, as well as the reduction in transport-related CO2 emissions. The model has been used for the analysis of Belgian transport data on palletized building materials for 2011 and collected within the framework of the ‘Build Over Water’ project. The results of this analysis emphasise a clear potential for a modal shift of these palletized building materials to the inland waterways, substantiating the previous results of the feasibility analysis (VUB and COMiSOL, 2006). The optimal locations and the potential for modal shift also depend on the chosen number of RWDCs. Therefore, two concepts have been used: ‘concept 30 km’, where the aim is to minimize the number of RDWCs; and ‘concept 15 km’, where the RWDCs are seen as small transhipment platforms. The latter permits a greater number of RWDCs, which is interesting as it lowers both the pre- and post-haulage distances, and the captured volume. For this analysis, the optimal number of RWDCs is 9 for ‘concept 30 km’ and 27 for ‘concept 15 km’. With 9 optimal RWDCs, over 300,000 tons of palletized building materials can be transhipped to the inland waterways at a profitable cost; in ‘concept 15 km’, this amount rises to almost 450,000 tons. The profitable tonnages increase by approximately 15% when administrative cost savings (0.5€/ton) or a general road pricing system (0.15€/ton) are taken into account. However, the introduction of depot costs reduces the profitability of the concept. Overall, the crucial factor is the amount of kilometres for pre- and post-haulage. Besides cost efficiency, the modal shift of palletized goods has also environmental and societal benefits. For the moment, only the CO2 emissions are included in the external cost analysis, but their results already illustrate the ecological importance of the modal shift. It has to be noted that the data used captures just a small part of the overall potential of transported palletized goods. The potential in the building sector is estimated at 6–7 million tons. Additionally, several fast moving consumer goods have shown potential too. Another limitation is the lack of detailed and international data. Coherent data collection is often limited to national/regional borders, whereas transport networks and transport flows are not. In the case of intermodal transport in particular – for which the break-even distance is an important concept – international flows represent a big potential, but this has not yet been included in the analysis. Further research will focus on the broadening of the geographical scale of the model. Parallel to this, the model will be updated to the latest markets and cost evolutions. Additionally, it is the aim to include other land use and traffic factors into the location analyses.
Acknowledgements The authors are thankful for the financial support from ‘Research Centre Mobilo on commodity and passenger flows’ and VIM (Vlaams Instituut voor Mobiliteit) for this research.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
References Algemene Directie Statistiek en Economische informatie (ADSEI), 2010. [General Agency of Statistics and Economic Information], Belgium Federal Government, FOD Economie, KMO, Middenstand en Energie, Brussels. [FOD Economy, SME, Private Sector and Energy]. Arnold, P., Peeters, D., Thomas, I., Marchand, H., 2001. Pour une localisation optimale des centres de transbordement intermodaux entre réseaux de transport: formulation et extensions [For an optimal localisation of intermodal transshipment hubs between transport networks: formulation and extensions]. Le Géographe Canadien 45 (3), 427–436. Aykin, T., 1995. The hub location and routing problem. European Journal of Operational Research 83, 200–219. Blauwens, G., Meersman, H., Sys, C., Van de Voorde, E., Vanelslander, T., 2011. Kilometerheffing in Vlaanderen: De impact op havenconcurrentie en logistiek. Steunpunt goederenstromen [Road Pricing in Flanders: The Impact on Ports Competition and Logistsics]. (Legal Depot Number D/2011/11.528/1). Antwerp. CE Delft, INFRAS, Fraunhofer ISI, 2011. External Costs of Transport in Europe: Update Study for 2008 (Report September 2011). CE Delft, Delft, Netherlands. Church, R.L., Sorensen, P., 1994. Integrating Normative Location Models into GIS: Problems and Prospects with the p-Median Model (Technical Report 94-5). University of California, Santa Barbara, National Center for Geographic Information and Analysis. Cornillie, I., Macharis, C., 2006. Palletvervoer via de binnenvaart, een haalbaar alternatief? [Pallettransport via the inland waterways, a feasible alternative?]. In: Despontin, M., Macharis, C. (Eds.), Mobiliteit en (groot)stedenbeleid. VUBPress, pp. 207–238. Daganzo, C.F., 2005. Logistics System Analysis, fourth ed. Springer, Heidelberg, p. 296. Danielis, R., Marcucci, E., Rotaris, L., 2005. Logistics managers stated preferences for freight service attributes. Transportation Research Part E 41 (3), 201–215. De Munck, L., 2010. Palletvervoer via de binnenvaart: een business met toekomst. [Pallettransport via the Inland Waterways: A Business of the Future]. Provinvial Academy Limburg, Log-IC, Hasselt. den Boer, L.C., Otten, M., van Essen, H., 2011, STREAM International Freight 2011: Comparison of Various Transport Modes on a EU Scale with the STREAM Database. CE Delft, Delft, Netherlands. Dijkstra, E.W., 1959. A note on two problems in connexion with graphs. Numerische Mathematik 1, 269–271. ESRI, 2010. ARCMAP 10. Network Analyst Users Guide. Essenciál Supply Chain Architects, 2011. Comparative Quantified Cost Analysis between Inland Navigation and Road Transport, for Palletised Goods. N554/ 2010 – Belgium. Preliminary Study for Waterwegen en Zeekanaal NV and NV De Scheepvaart. European Commission, 2011. EU Transport in Figures: Statistical Pocketbook 2011. Publications Office of the European Union, Luxembourg. Freight Best Practice, 2005. Key Performance Indicators for the Pallet Sector: Benchmarking Guide. Department for Transport, London. Groothedde, B., Ruijgrok, C., Tavasszy, L., 2005. Towards collaborative, intermodal hub networks: a case study in the fast moving consumer goods market. Transport Research Part E 41, 567–583. Hillsman, E.L., 1984. The p-median structure as a unified linear model for location– allocation analysis. Environment and Planning A 16 (3), 305–318. Kayikci, Y., 2010. A conceptual model for intermodal freight logistics centre location decisions. Procedia Social and Behavioral Sciences 2, 6297–6311.
53
Kreutzberger, E., Macharis, C., Woxenius, J., 2006. Intermodal versus unimodal road freight transport – a review of comparisons of the external costs. In: Jourquin, B., Rietveld, P., Westin, L. (Eds.), Towards Better Performing Transport Systems. Taylor and Francis, London, pp. 17–42. LOGIQ, 2000. Final Report for Publication of LOGIQ Research Project (Contract No.: IN-97-SC.2237). Gruppo CLAS, Milan. Mommens, K., Macharis, C., 2012. Pallets on the Inland Waterways – Policy Paper (Legal Depot Number D/2012/11.528/13). Antwerp: Steunpunt Goederen- en Personenvervoer. Pekin, E., Macharis, C., Meers, D., Rietveld, P., 2012. Location analysis model for Belgian intermodal terminals: importance of the value of time in the intermodal transport chain. Computers in Industry 64, 113–120 (special issue). Poppink, P., 2005. Project roept vragen op: Distrivaart laat beperkingen modal shift zien [Project evokes some questions: Distrivaart shows the limits of the modal shift]. Transport en Logistiek 4, 26–27. Ricci, A., Black, I., 2005. The social costs of intermodal freight transport. Transport Economics 14, 245–285. Rutten, B.J.C.M., 1998. The design of a terminal network for intermodal transport. Transport Logistics 1 (4), 279–298. Service d’études techniques des routes et autoroutes (Sétra), 2008. Le transport fluvial de palettes à Paris: une activité peu ordinaire [Barge transport of pallets in Paris: a rare activity]. Revue Transports. May 2008 (Nr. 5). Ministère de l’écologie, du développement et de l’aménagement durables, Paris. Teitz, M.B., Bart, P., 1968. Heuristic methods for estimating the generalized vertex median of a weighted graph. Operations Research 16, 955–961. Van Dorsser, J.C.M., 2004. Economische Haalbaarheid van Palletvaart [Economic Feasibility of Pallettransport via the Inland Waterways]. Master Thesis, Erasmus University, Rotterdam. Van Lier, T., Macharis, C., 2013. Emissions of barge transport: a life-cycle assessment approach. In: Proceedings of 13th World Conference on Transport Research. Rio de Janeiro, Brazil, 15–18 July 2013 (to appear in selected proceedings). Vannieuwenhuyse, B., Gelders, L., Pintelon, L., 2003. An online decision support system for transportation mode choice. Logistics Information Management 16 (2), 125–133. Verbeke, F., Cornillie, I., Macharis, C., 2007. De derde golf in de binnenvaart: palletvervoer succesvol getest in de praktijk [The third wave in barge-transport: pallettransport successfully tested in practice]. In: Vervoerslogistieke Werkdagen 2007. Nautilus Academic Books, Grobbendonk, Belgium, pp. 376– 389. Verbeke, F., Macharis, C., Mommens, K., 2012. Shifting pallets to the inland waterways: the combination of theory and practice. In: Weijers, S., Dullaert, W. (Eds.), Bijdragen Vervoerslogistieke werkdagen, 29–30 November. Zelzate University Press, Venlo, pp. 331–344. Vlaams Instituur voor Mobiliteit (VIM), 2012a. De voornaamste resultaten van de proefvaarten van ‘‘Build Over Water’’ [The Main Results of the Test Bargetransports of the ‘‘Build Over Water’’]. VIM, Diepenbeek. Vlaams Instituut voor Mobiliteit (VIM), 2012b. Eindrapport Modelproject en Distributieanalyse Build over Water [Final Report of the Project and Distribution-analysis Build Over Water] (Project Nr. P029). VIM, Diepenbeek. Vrije Universiteit Brussel (VUB), COMiSOL, 2006. Haalbaarheidsstudie voor de concrete implementatie van de binnenvaart voor vervoer van pallets en de daarmee verbonden stadsbevoorrading [Feasibility-analysis for the Concrete Implementation of Pallettransport via the Inland Waterways and the Related City Distribution]. Brussels.
Contents lists available at ScienceDirect
Journal of Transport Geography journal homepage: www.elsevier.com/locate/jtrangeo
Location analysis for the modal shift of palletized building materials Koen Mommens ⇑, Cathy Macharis Vrije Universiteit Brussel, MOBI, Pleinlaan 2, Brussels 1050, Belgium
a r t i c l e
i n f o
Keywords: Location analysis model Pallets Intermodal transport
a b s t r a c t At present, the distribution of palletized building materials is mostly carried out by trucks, despite their movements having negative effects on society, the economy and the environment. However, these problems can be reduced if the transport of palletized goods is shifted to inland waterways. By doing so, the goods are bundled for the main haulage by barge. In order to reduce the transport distances by truck to an absolute minimum, a possible last-mile distribution would have to be organized via a limited number of directly canal-served hubs. The locations of those hubs are crucial for the feasibility of modal shift. This study advances the transport geography literature by elaborating a location analysis model specifically for palletized goods. This model determines the optimal hub location by taking into account the large variation of origins and destinations of transport flows, while the introduction of a cost structure enables potential economic gains (cost savings) and reductions in CO2 emissions to be calculated. The analysis is performed for transport data on palletized building materials in Belgium. Two concepts were defined, which resulted in an optimal intermodal network of 9 hubs and one with 27 hubs; through the implementation of these networks, respectively 26% and 38% of the transport flows can be shifted to the inland waterways at a profitable cost. It can be expected that over time these percentages will increase further. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Most transport and mobility movements are performed on the road network (European Commission, 2011). Consequently, they are responsible for several negative externalities, such as congestion, pollution, noise nuisance and accidents. These externalities affect our daily lives, our economy and the environment. Other transport modes – like barge and rail – create similarly negative externalities, but in significantly lower proportions than road transport (CE Delft et al., 2011). As the demand for transport and mobility is expected to continue growing in the coming years and decennia, it is necessary to take into account those other transport modes as reasonable alternatives to the dominant road transport. For several Western European countries, inland waterways offer many opportunities, as its network is wide and mostly underutilised. In the past, large numbers of merchant transport operated within this network, with all kinds of goods transported by barge to the inner centre of Western European cities. In later ages, this practice disappeared, and canals were closed in many cities. However, following the containerisation of maritime transport that started in the 1970s, inland waterways were rediscovered in the 1990s by shippers, logistics service providers and ⇑ Corresponding author. E-mail addresses: [email protected] (K. Mommens), cathy.macharis@ vub.ac.be (C. Macharis). 0966-6923/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jtrangeo.2013.11.001
governments. After big bulk (as traditional first wave) and containers (as second wave) (VUB and COMiSOL, 2006), the challenge now is to find new types of loading units which can be transhipped to the waterway network. Palletized goods constitute one possibility in this regard. Road transport of palletized goods is currently carried out at a very competitive price, especially when the external costs are not included (Ricci and Black, 2005; Van Dorsser, 2004). However, some initiatives of transporting palletized goods via inland waterways have emerged in recent decades throughout Western Europe. During the early 2000s in the Netherlands, palletized drinks were transported via inland waterways. This Distrivaart project was abandoned in 2004, as the commercial basis was too small (Poppink, 2005). In another project in the Netherlands, the urban distribution of different kind of goods, including palletized goods, is organized by Mokum Mariteam via the local waterway network of Amsterdam. In the region of Paris (France), palletized building materials have been transported by barges since 1987 (Sétra, 2008). In Belgium also, the initial focus was oriented towards the construction sector, as a feasibility analysis (VUB and COMiSOL, 2006) indicated a clear potential for a modal shift of these goods. As a result of this study, different practical experiments were conducted throughout Flanders (Verbeke et al., 2007; VIM, 2012a,b). Currently, transport of palletized building materials on seven fixed routes connecting suppliers and customers that are directly adjacent to canals are being shifted to inland waterways.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
In order to organize the transport of palletized goods by barges in a larger and more (cost-) effective way, a network of transhipment hubs – or regional water-bound distribution centres (RWDCs) – needs to be created within the supply chain. Palletized goods can be bundled onto barges at those hubs, and more sustainable first- and last-mile distribution can be organized. In this way travel distances by trucks can be reduced to an absolute minimum. Previous theoretical studies (Cornillie and Macharis, 2006; Groothedde et al., 2005; Van Dorsser, 2004; VUB and COMiSOL, 2006) and practical experiments (Poppink, 2005; Verbeke et al., 2007, 2012; VIM, 2012a,b) have demonstrated that the length of both pre- and post-haulages has a large impact on the economic feasibility of the modal shift of palletized goods. The hub locations determine the lengths of both pre- and post-haulages: the closer the transhipment hubs are to the important volumes, the lower the number of pre- and post-haulage tonkm that needs to be performed by trucks. As a result, the economic, societal and ecological costs incurred from vehicular travel can be reduced. The need for a substantiated identification of the transhipment locations thus arises. Additionally, both private and public sectors are interested in the economic and ecologic potential of an implementation of the concept. In order to answer to the above-stated needs, we have developed LAMBTOP (Location Analysis Model for Barge Transport of Pallets). The model is constructed for the Belgian territory (Mommens and Macharis, 2012). This paper is structured as follows: in Section 2, we describe the challenges and principles of a modal shift of palletized goods via a network of RWDCs; the developed methodology is explained in Section 3. The obtained results based on data on palletized building materials transported within Belgium in 2011 are discussed in Section 4. We end with the conclusions. 2. Hub concept The integration of a main haulage by barge implies that suppliers and/or customers who are not located near an inland waterway need an initial and/or final haulage via road. In those cases, one can talk about intermodal transport. Still, the supply chain of palletized goods has very different logistical characteristics in comparison with inland waterway transport (IWT) in the form of bulk and containers. Table 1 illustrates these characteristics so as to indicate the challenges that must be dealt with when shifting palletized goods to the inland waterways. In order to match the characteristics of the palletized goods’ supply chains with those of the typical inland waterway transport, well-chosen transhipment hubs must be created within the supply chain. These hubs (or RWDCs) work as regional distribution centres whereby flows of palletized goods are bundled and transhipped to a barge. RWDCs can be supplied via road and inland waterways. Fig. 1 illustrates the possible flows of the supplier’s and/or customer’s warehouse, depending on the possible directly canal-served facilities:
Table 1 Supply chain characteristics of different cargo types. Source: Verbeke et al. (2012). Supply chain characteristic
Bulk
Containers
Pallets
Typical IWT
Number of SKU (stock keeping unit) Volume per SKU Speed of delivery
Few
No issue
Many
Few
High Low
No issue High
High Very low
Number of drops
Low
Low
Low Very high High
Very low
45
1. Directly canal-served supplier’s warehouse ) inland waterway ) RWDC ) post-haulage via road ) non-canal-served customer’s warehouse. 2. Directly canal-served supplier’s warehouse ) inland waterway ) directly canal-served customer’s warehouse. 3. Non-canal-served supplier’s warehouse ) pre-haulage via road ) RWDC ) inland waterway ) RWDC ) post-haulage via road ) non-canal-served customer’s warehouse. 4. Non-canal-served supplier’s warehouse ) pre-haulage via road ) RWDC ) inland waterway ) directly canal-served customer’s warehouse. 3. LAMBTOP methodology 3.1. Introduction Since in reality not every supplier or customer of palletized goods has a location near an inland waterway, a network of RWDCs needs to be created (VUB and COMiSOL, 2006). The optimal location of these RWDCs is crucial for all stakeholders, as these locations have a large impact on the profitability of the intermodal transports. Consequently, their locations will also define the potential turnover of the RWDC (Arnold et al., 2001; Aykin, 1995; Kayikci, 2010). The LAMBTOP (Location Analysis Model for Barge Transport Of Pallets) that will be discussed in this paper was developed to determine the optimal location of RWDCs in Belgium. Besides their locations, the model calculates the financial cost of the modal shift and the potential turnover for each distribution centre. This is important information given that transport costs are one of the main determinants in the intermodal decision-making process (Danielis et al., 2005; LOGIQ, 2000; Vannieuwenhuyse et al., 2003). Moreover, the transport price has proved itself as a clear bottleneck in previous studies, not to mention in the Distrivaart project (Poppink, 2005; VIM, 2012a; VUB and COMiSOL, 2006). In addition to the financial outcome, the model also calculates the potential reduction in CO2 emissions, so as to illustrate the ecological benefit of the modal shift. We focus on the financial cost structure in Section 3.2 before explaining the used methodology for CO2 comparison in Section 3.3. The methodology of the LAMBTOP is summarized in Section 3.4. Finally, we focus on the data collection in Section 3.5. 3.2. Cost comparison The calculation of the financial difference between unimodal road transport and intermodal transport is performed with a cost structure based both on theoretical analyses (Essenciál Supply Chain Architects, 2011; Freight Best Practices, 2005) and practical experiments with palletized building materials (De Munck, 2010; Verbeke et al., 2007; VIM, 2012b). Information regarding these experiments was obtained through contact with several transport experts1 that accompanied these field tests. In the first stage, the cost structure only considers direct transport-related costs. In subsequent stages, different scenarios have been added, allowing us to test the sensitivity of the analysis. Firstly, additional depot costs ranging from 1€/ton to 4€/ton have been introduced into the intermodal supply chain. A second scenario takes into account administrative savings accrued through saved transport documents, payment transactions and invoices. The last scenario tests the impact of introducing a general road
1 The transport experts assist Flemish companies in their search for possibilities to shift (portions of) their transport flows to the inland waterways. The advice and services provided to the transport companies by these transport experts are free of charge, as the experts are appointed by a cooperation of waterway administrators and enterprise unions.
46
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
Fig. 1. Supply chain of the transport of palletized goods by barge. Source: VUB and COMiSOL (2006).
pricing system conform the rate of the LWK-Maut in Germany (0.15€/km) (Blauwens et al., 2011). Management costs, investment costs and external costs (congestion, accidents, emissions, etc.) are not included. The financial costs are expressed in €/ton, whereas the distance dependent variables are expressed in €/tonkm. The time trade-offs created by the intermodal transport are not included. This is justified, as the construction sector’s interest and willingness in a modal shift of their flows prove that any gains in reliability and cost efficiency brought about by this shift are far more important than any created time trade-offs. The following cost structure is obtained for the supply chain of the unimodal road transport:
C R ¼ C LT þ C H dr þ C UT
ð1Þ
Under the assumption that both the supplier’s and the customer’s warehouses have no directly canal-served facilities, the cost structure for the intermodal transport’s supply chain, which combines the costs of the different steps, is defined as follows:
C I ¼ C LT þ C PH dpr þ C UT þ C LB þ C B db þ C UB þ C LT þ C PH dpo þ C UT
ð2Þ
The difference between the cost of the intermodal transport (CI) and the unimodal road transport (CR) is illustrated in Fig. 2, representing the basic structure of the cost comparison. Rate books reveal that transportation costs mainly depend on distance.
Fig. 2. Cost structure. Source: own composition.
Moreover, the relationship between those costs and distance is approximated adequately by using a linear increasing function (Daganzo, 2005). Both transports start with an initial cost of loading the truck (CLT). In the intermodal variant, the truck drives over a distance (dpr) to the optimal RWDC at a higher cost (CPH) than the cost of the main haulage by truck of the unimodal transport (CH). This is due to the assumption of a probable empty return haulage in the case of pre- and post-haulages (100% empty kilometres) (De Munck, 2010; Essenciál Supply Chain Architects, 2011), whereas the main haulages of the unimodal transport are assumed to be done at a ratio of 26.5% empty kilometres (Freight Best Practices, 2005). Once the truck arrives at the RWDC, it has to be unloaded (CUT), after which the palletized goods have to be loaded into a barge (CLB). The loading of the barge is assumed to be done by forklifts or mobile cranes. Both techniques have proved themselves as the most efficient during practical experiments where different loading techniques were tested (Verbeke et al., 2012; VIM, 2012b). Once the barge is filled up with palletized building materials to an average loading factor (weight) of 95%, it navigates to an average cost (CB) over a distance (db). Additionally, the assumption is made that the barge transport counts 24% empty kilometres. This percentage has been validated by a market player. Upon arrival at the second RWDC, the barge is unloaded (CUB) using the same techniques as when it was loaded. The palletized goods are then transhipped onto a truck (CLT), which brings them to the final destination – as we assume – at the same cost as the pre-haulage (CPH). The additional transhipment costs at the RWDC (CUT, CLB, CUB and CLT) have a large impact on the overall price of the intermodal transport. For the combination of one truck (un)loading (CUT) and one barge (un)loading (CLB), a break-even distance of 60.8 km is needed. The break-even distance is the distance at which the costs of intermodal transport equal the costs of unimodal road transport (Pekin et al., 2012; Rutten, 1998). In cases where both pre- and post-haulages are needed, the break-even distance for the transhipment costs made at the RWDC is 121.7 km. The costs of the pre- and post-haulages are not included in these break-even distances. One cannot calculate a general break-even distance, simply because it depends on the length of the pre- and/or post-haulages, which varies for every origin–destination combination.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
In a comparison of the main haulages, the financial cost of transportation by barge (CB) is more than half the financial cost of main haulage by truck (CH). The financial profit from an intermodal shift should thus be made in this section of the supply chain in order to compensate for the extra handlings at the RWDC and possible pre- and/or post-haulages (see Fig. 2). Although, as stated before, no general break-even distance can be calculated, several scenarios can be analysed for different scenarios with different pre- and/or post-haulages distances (see Fig. 3). The initial cost contains the costs of the pre- and post-haulages and all the transhipment costs of both barge and truck. As such, the break-even distance can be calculated for each scenario. In the case where both the supplier’s and customer’s warehouse have directly canal-served facilities – and consequently no preand post-haulage, and no additional transhipment costs of (un)loading the truck are needed – choosing the intermodal alternative is theoretically always a profitable one. If one pre- or post-haulage is needed, the extra initial costs of these haulages and loading and unloading of the truck must be taken into account. The break-even distance of the first scenario (with 5 km of pre- or post-haulage) is 68.1 km. In cases with a pre- or post-haulage of 30 km, the break-even distance rises to 140.4 km. The break-even distance of the minimum scenario of pre- and post-haulage (5 km) is 136.2 km. The maximum route distance via road between Belgian municipalities located within a buffer of 30 km of an inland waterway is approximately 250 km. Within this distance, all stated intermodal scenarios (up to 40 km of preand post-haulages) are profitable. More precisely, the tipping point for a break-even distance of 250 km is 46.8 km of pre- and posthaulage. It should be noted that in reality the distance of the main haulage by barge is often much longer than the main haulage by truck. This is largely due to the difference in density between the inland waterway network and road network. In the Belgian case, the difference in length of the main-haulage by barge and by unimodal truck is approximately 25% on average, which implies a decrease in the so-called tipping point to a more realistic 28.8 km. This distance corresponds to the 30 km mentioned in previous studies as the maximum distance of pre- and post-haulage for which the modal shift of palletized goods can be profitable (Cornillie and Macharis, 2006; Essenciál Supply Chain Architects, 2011; Poppink, 2005). 3.3. CO2 comparison In general, intermodal transport creates lower external costs (pollution, emissions, noise, congestion and accidents) than
47
unimodal road transport (CE Delft et al., 2011; Kreutzberger et al., 2006). In this paper, only the transport-related CO2 emissions are taken into account. Some assumptions have to be made in order to calculate the CO2 emissions. First of all, as the obtained data lacks information on vehicle type, the haulages via road are assumed to be done by a truck-trailer combination (36.1 ton average gross vehicle weight; 26.2 ton loading capacity). The CO2 emissions are based on European average figures (10% urban, 15% non-urban and 75% motorway). The study we used – the STREAM International Freight study (Den Boer et al., 2011) – focuses on different transport modes (rail, road, inland waterways and short sea shipping). The information on average truckloads enables us to calculate the CO2 emissions for all the haulages by truck (unimodal variant, and pre- and post-haulage). The pre- and post-haulages are assumed to have an empty return haulage (100% empty kilometres), while the main haulage in the unimodal variant is assumed to have 26.5% empty kilometres (Freight Best Practices, 2005). The CO2 emissions for barge transport are calculated on the bases of the Flemish Environmental Agency’s (Vlaamse Milieu Maatschappij (VMM)) calculations, using the EMMOSS model of Transport & Mobility Leuven (Van Lier and Macharis, 2013). Moreover, an assumption is made that transport by barge is carried out with an average load factor (weight) of 95%, and an average empty running of 24%. 3.4. Methodology The LAMBTOP is a geographic information system (GIS) based model, which consists of different network-layers, each representing a transport mode (road and barge). The locations of departures and destinations are connected to the network layers by their corresponding nodes. In many cases, the municipality centres – defined as the main church of the municipality – act as those nodes (Fig. 4). If it is known that a departure and/or destination location is directly canal-served, this is included as such in the analysis. The network for Belgium is built by combining the following digital databases: The inland waterways layer is extracted from the Environmental Systems Research Institute (ESRI) data set for Europe. The road layer and municipality layer are obtained from the MultiNet database of Tele Atlas. The model is set up for Belgium, but can easily be transformed for other regions, and preferably expanded to a European scale.
Fig. 3. Cost comparison of unimodal road transport versus intermodal transport (IT) scenarios. Source: own composition.
48
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
Fig. 4. The network in ArcGis. Source: own composition.
As a first step, the obtained geographical information about the flows of the palletized goods is uploaded to the model in the form of an origin–destination matrix (OD-matrix) (Fig. 5). These ODcombinations are linked to their corresponding nodes in the origin–destination-point layer. It enables us to identify and map the transport flows of palletized goods. The model identifies the OD-combinations and computes the unimodal routes travelled via the road network. These routes are calculated by a shortest time path algorithm; Dijkstra’s (1959) algorithm is used for every path algorithm in the entire analysis (unimodal routes, buffer, location analysis, intermodal routes). The second step is the distribution analysis of the transport of palletized goods, which will determine the future locations of the RWDCs. The distribution locations (the nodes of the departure and arrival locations) are weighted by the sum of the tonnage of
Fig. 5. The structure of the LAMBTOP. Source: own composition.
all the routes that start and arrive in these receptive nodes. Next, a precondition is formulated; the distribution locations used as the ‘market area’ for the location analysis of the RWDCs are limited to the nodes which are located within a predefined buffer of an inland waterway (illustrated in Fig. 6). The buffer is fixed once at 15 km and once at 30 km within this analysis, using the road network for a shortest time path algorithm. Although the critical maximum distance of the pre- and post-haulage necessary for the overall intermodal transport to be profitable depends on the overall transport distance and the used cost structure, it can be assumed that 30 km of pre- and post-haulage is the ultimate maximum for the Belgian case (Cornillie and Macharis, 2006; Essenciál Supply Chain Architects, 2011; Poppink, 2005; Section 3.1). By using this delimitation, intermodal routes with too long and consequently too expensive pre- and post-haulages will be excluded. The potential locations of the RWDCs are defined as locations on an inland waterway of minimal class III (P600 ton), and lying within a predefined distance (50 m) of a trafficable road. Thanks to this precondition, future RWDCs will not be located in a pedestrian city centre or in a protected nature reserve. Furthermore, RWDCs will already have a direct connection to the existing road network, thereby avoiding heavy investments in road infrastructure. The determination of the optimal locations (step 3) is based on the ‘Location–Allocation’ procedure of the ‘ArcGIS Network Analyst’ tool. The procedure starts with the calculation of the shortest path between every distribution location and every potential RWDC location using the road network and the algorithm of Dijkstra (1959). Then an edited version of the obtained cost matrix is constructed (Hillsman, 1984), which enables the heuristic to solve a variety of different problem types. Next, the location–allocation process generates a set of semi-randomized solutions, before applying the vertex substitution heuristic of Teitz and Bart (1968) to create a group of good solutions (Church and Sorensen, 1994). A metaheuristic then combines these good solutions to find better solutions until no additional improvement is found. Finally the metaheuristic delivers the best solution found (ESRI, 2010). The optimal locations of the RWDCs vary with the number of chosen RWDCs. This number is chosen on the basis of the ‘market share’ and their spatial distribution. In the concept which uses a buffer of 30 km (‘concept 30 km’), the idea is that it is better to enlarge one directly canal-served distribution centre than open two of them in the same ‘market area’. Thus the number of RWDCs will be limited. In the concept with a buffer of 15 km (‘concept 15 km’), RWDCs are seen as small transhipment platforms. This permits a larger number of RWDCs, which is interesting because it lowers the pre- and post-haulage distances and the captured volume. Therefore, the hubs may need to be combined with other logistic activities to achieve required volumes.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
49
Fig. 6. Distribution analysis (tonnage per municipality) and potential market area. Source: own composition.
Once the number and the optimal location of the future RWDCs is set, a GIS network is created specifically for the intermodal transport. It combines the road layer (for the pre- and post-haulage), the RWDC locations (as transhipment nodes) and the inland waterway network layer. Within this newly created intermodal network, a shortest time path algorithm for the pre- and post-haulages and a shortest route algorithm for the main haulage by barge are performed for every OD-combination. The respective distances of the unimodal routes by road and the intermodal routes are calculated for all the OD combinations, and the distances are then linked to the cost structure. A cost analysis is performed for every individual OD-route. The combination of all routes gives a global overview in which the routes where the modal shift is cost-efficient will describe a realistic potential modal shift and a realistic potential turnover (in tons) of the future RWDCs. Assuming that these profitable transports will be shifted towards the inland waterways, it is possible to calculate the saved truck movements, and consequently an estimation of the potential CO2 reduction. Different cost scenarios – an introduction of road pricing, an introduction of depot costs (1€/ton to 4€/ton) and administrative cost savings – enable us to test the sensitivity of the analysis. 3.5. Data collection The analysis is performed for Belgian transport flows of palletized building materials for the year 2011. The construction sector
is an import economic motor for the country, representing 7–8% of Belgium’s GNP. The sector also has a large impact on mobility, given that construction-related transport represents 25% of freight transport on Belgian highways (VIM, 2012b) and a large portion of this freight is loaded on pallets. In total, almost 53 million tons of palletized goods are transported within Belgian borders each year (ADSEI, 2010). It is impossible to identify the palletized building materials from these last data, but a feasibility analysis (VUB and COMiSOL, 2006) demonstrated that 6–7 million tons of this type of good could be shifted to inland waterways. The used data for 2011 were collected in the ‘Build Over Water’ project (VIM, 2012a,b). The goal of this project was to analyse the feasibility of a modal shift of palletized building materials to inland waterways. This was done through a combination of different practical experiments, and by a distribution and location analysis on the basis of the LAMBTOP. The flows were obtained via a dedicated survey in which 9 out of the 50 reported producers participated. Together, they represent 1.163 million tons of transported palletized goods; in other words, the survey covers approximately 1/6 of the total volume that could be shifted to inland waterways. The data contain information about the producer, the tonnage, and the origin and destination at the spatial scale of the municipality. For the directly canal-served sites, the address location is used as the geographic node. For all non-canal-served locations, or cases for which no address information is given, the tonnages were assigned to the municipality centre. In cases where a municipality contains both water-served and non-canal-served destinations,
50
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
Table 2 Overall potential turnover of RWDCs. Source: own composition. Concept 30
Concept 15
Location
Volume (ton)
Location
Volume (ton)
Location
Volume (ton)
Location
Volume (ton)
Kortrijk Antwerp Hasselt Aalst Mol Leuven Pont-à-Celles Bruges Liège
346,988 197,149 142,807 101,526 88,261 77,109 66,841 62,918 58,773
Wervik Waregem Dilsen-Stokkem Hasselt Beringen Willebroek Beerse Liège Temse
160,614 156,500 65,210 64,330 63,132 59,531 53,283 47,482 41,998
Riemst Leuven Herentals Antwerp Bocholt Halle Brussels Aalst Roeselare
37,131 36,600 36,446 29,106 29,013 28,081 27,452 25,711 25,710
Charleroi Jabbeke Gent Pecq Manage Aalter Lier Huy Saint Ghislain
23,951 23,398 23,305 18,381 17,273 11,331 10,213 3917 3766
the assigned tonnage is divided proportionally between these locations.
4. Results and discussion The 1.163 million tons of palletized building materials creates an origin–destination-matrix (OD-matrix) of 1462 combinations. This matrix was uploaded into the model and the distribution analysis is illustrated in Fig. 6. The production locations are – as origins – mainly concentrated in the west and north of the country. The tonnage of the destinations is more equally distributed, but some general larger concentrations can be distinguished in the Flemish
part of the country, the ABC axis (Antwerp–Brussels–Charleroi) and around the city of Liège. In order to limit the pre- and/or post-haulage distances – in prospect of the profitability of the modal shift – only the distribution locations which lie within the predefined buffer of 30 km (in the left map) and 15 km (in the right map) of an inland waterway are selected as the ‘market area’ for future optimal RWDCs. For the concept with a buffer of 30 km, the location analysis works out an optimal number of nine RWDCs. The buffers of 30 km around these nine optimal RWDC locations cover 96.5% of the 1.184 million tons which have been defined as the ‘market area’. A market area with less than nine RWDCs lacks the ability to serve considerable market areas within a reasonable pre- and post-haulage distance. Once
Fig. 7. Overall potential turnover of RWDCs and the intermodal routes. Source: own composition.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
51
Fig. 8. Volume and CO2 emissions savings for the profitable intermodal routes of each scenario. Source: own composition.
Fig. 9. Profitable intermodal routes for scenario with road pricing and depot cost (left: ‘concept 30 km’; right: ‘concept 15 km’). Source: own composition.
more than nine RWDCs are chosen, the added value of the new RWDCs becomes negligible. In the case of a market area of 15 km, the model returns 27 optimal RWDC locations. They cover 99.2% of the 1.122 million non-canal-served tons distributed within the market area. Four of the 27 optimal locations are closely located to a directly canal-served origin location. It can be expected that, in case of implementation, these neighbouring locations will merge in time to one transhipment location. This is an illustration that the optimal locations in practice will be adjusted to field characteristics, because currently – except for the proximity of a trafficable road – the location analysis does not take into account other land use and traffic factors. Public and private sector are aware of that, and consequently they are not only focussing on the exact
optimal locations, but if necessary also to suitable area’s near those optimal locations. Fig. 7 illustrates the spatial distribution of the optimal RDWC locations for both concepts (left: ‘concept 30 km’, right: ‘concept 15 km’). Additionally, the intermodal routes and the overall potential turnover (in tons) of the RWDCs are shown in Table 2. The overall potential turnover of a RWDC is the tonnage which would be transhipped there if all given OD-combinations are shifted to the inland waterway. Of course, this is not a realistic assumption because, according to the cost structure, the flows have to cope with a break-even distance (Section 3.2). The volumes per RWDC for the ‘concept 15 km’ are low, as they are seen as small transhipment platforms which may or may not be combined with other lo-
52
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
gistic activities to achieve required volumes. When applying the cost structure to the intermodal and unimodal routes, it can be seen in Fig. 8, that today over 300,000 tons of palletized building materials can be transhipped at a profitable cost via the 9 optimal RWDC locations to inland waterways. In the case of the 27 RWDCs (or ‘concept 15 km’), the profitable amount rises to almost 450,000 tons. These data are for the survey which covered approximately 1/6 of the total potential transport flows of palletized building materials. Therefore, it can be assumed that the real-life modal shift of palletized building materials will be larger than those stated tonnages. The potential savings in transport-related CO2 emissions are shown in Fig. 8. The total CO2 emissions savings for the profitable intermodal routes with respect to their unimodal alternatives varies between 232 and 922 tons, depending on the concept and scenario. The proportional ecological gains are clearly higher in the ‘concept 15 km’ scenario. This is also the case for the scenario where a general road pricing system that conforms to the rate of the LWK-Maut in Germany is implemented (Blauwens et al., 2011). For this scenario, the profitable modal shift increases by approximately 15%: to over 350,000 tons for ‘concept 30 km’ and almost 520,000 tons for ‘concept 15 km’. The introduction of administrative costs savings to the general cost structure has more or less the same impact as road pricing. When depot activity costs are added to the general cost structure, the profitability of the concept is reduced and added value needs to be created for depot activities. Added value can be created through additional services at the RWDC (VIM, 2012b) or through consolidation with other types of cargo, especially in the construction sector, where the industry is often located near the inland waterways and where bulk materials have a similar distribution pattern as the palletized cargo. The RWDC could thus also serve as a transhipment hub for bulk materials. Combining the transport of bulk and palletized cargo in one barge is, however, not suitable, as it enlarges the transhipment costs. Finally, the added depot activity costs can also be encountered by costs savings through consolidation of different deliveries into one post-haulage. The last scenario combines the road pricing scenario with a depot cost of 2€. Under these circumstances, 17.8% and 20.2% of the initial 1.163 million tons can be shifted at a profitable cost for ‘concept 30 km’ and ‘concept 15 km’ respectively. When mapping the intermodal routes of these profitable combinations (Fig. 9), it can be seen that they are mostly allocated to RWDCs in the west and the east of the country. Transport distances are therefore high, so the break-even distance is reached (Fig. 3). This fact advocates the introduction of an international RWDC network where average transport distances will be higher. Consequently, the financial and ecological benefits of the concept increases.
5. Conclusions The transport of palletized goods via Belgian inland waterways has evolved from a theoretical feasibility analysis (VUB and COMiSOL, 2006), which illustrated a clear potential for palletized building materials and fast moving consumer goods, to practical experiments and even different operational schemes for the transportation of palletized building materials. This multiplicity of experience has enabled the development in this paper of a cost structure based on transport-related financial costs. Different scenarios have been developed, including a general introduction of road pricing, administrative savings and depot costs. The cost structure shows an absolute economical potential for the modal shift of palletized goods transportation when both producer and customer are located near inland waterways. In cases where the supplier’s and/or customer’s warehouse is not located
near an inland waterway, RWDCs and pre- and/or post-haulages become inevitable. It is obvious that shorter haulage distances increase the (financial) feasibility of modal shift for palletized goods. So, the location choice of the RWDC, and consequently the minimization of pre- and/or post-haulages, is crucial for the success of the concept. In this paper, the LAMBTOP has been described. This model allows us to analyse the optimal RWDC locations. It has been developed to facilitate the modal shift via a network of hubs, and this for a new type of cargo (palletized goods) with specific characteristics. Moreover, the model takes into account the variation in origins and destinations to calculate these optimal locations. The model is also able to calculate the financial costs of the modal shift and the potential turnover of every RWDC, as well as the reduction in transport-related CO2 emissions. The model has been used for the analysis of Belgian transport data on palletized building materials for 2011 and collected within the framework of the ‘Build Over Water’ project. The results of this analysis emphasise a clear potential for a modal shift of these palletized building materials to the inland waterways, substantiating the previous results of the feasibility analysis (VUB and COMiSOL, 2006). The optimal locations and the potential for modal shift also depend on the chosen number of RWDCs. Therefore, two concepts have been used: ‘concept 30 km’, where the aim is to minimize the number of RDWCs; and ‘concept 15 km’, where the RWDCs are seen as small transhipment platforms. The latter permits a greater number of RWDCs, which is interesting as it lowers both the pre- and post-haulage distances, and the captured volume. For this analysis, the optimal number of RWDCs is 9 for ‘concept 30 km’ and 27 for ‘concept 15 km’. With 9 optimal RWDCs, over 300,000 tons of palletized building materials can be transhipped to the inland waterways at a profitable cost; in ‘concept 15 km’, this amount rises to almost 450,000 tons. The profitable tonnages increase by approximately 15% when administrative cost savings (0.5€/ton) or a general road pricing system (0.15€/ton) are taken into account. However, the introduction of depot costs reduces the profitability of the concept. Overall, the crucial factor is the amount of kilometres for pre- and post-haulage. Besides cost efficiency, the modal shift of palletized goods has also environmental and societal benefits. For the moment, only the CO2 emissions are included in the external cost analysis, but their results already illustrate the ecological importance of the modal shift. It has to be noted that the data used captures just a small part of the overall potential of transported palletized goods. The potential in the building sector is estimated at 6–7 million tons. Additionally, several fast moving consumer goods have shown potential too. Another limitation is the lack of detailed and international data. Coherent data collection is often limited to national/regional borders, whereas transport networks and transport flows are not. In the case of intermodal transport in particular – for which the break-even distance is an important concept – international flows represent a big potential, but this has not yet been included in the analysis. Further research will focus on the broadening of the geographical scale of the model. Parallel to this, the model will be updated to the latest markets and cost evolutions. Additionally, it is the aim to include other land use and traffic factors into the location analyses.
Acknowledgements The authors are thankful for the financial support from ‘Research Centre Mobilo on commodity and passenger flows’ and VIM (Vlaams Instituut voor Mobiliteit) for this research.
K. Mommens, C. Macharis / Journal of Transport Geography 34 (2014) 44–53
References Algemene Directie Statistiek en Economische informatie (ADSEI), 2010. [General Agency of Statistics and Economic Information], Belgium Federal Government, FOD Economie, KMO, Middenstand en Energie, Brussels. [FOD Economy, SME, Private Sector and Energy]. Arnold, P., Peeters, D., Thomas, I., Marchand, H., 2001. Pour une localisation optimale des centres de transbordement intermodaux entre réseaux de transport: formulation et extensions [For an optimal localisation of intermodal transshipment hubs between transport networks: formulation and extensions]. Le Géographe Canadien 45 (3), 427–436. Aykin, T., 1995. The hub location and routing problem. European Journal of Operational Research 83, 200–219. Blauwens, G., Meersman, H., Sys, C., Van de Voorde, E., Vanelslander, T., 2011. Kilometerheffing in Vlaanderen: De impact op havenconcurrentie en logistiek. Steunpunt goederenstromen [Road Pricing in Flanders: The Impact on Ports Competition and Logistsics]. (Legal Depot Number D/2011/11.528/1). Antwerp. CE Delft, INFRAS, Fraunhofer ISI, 2011. External Costs of Transport in Europe: Update Study for 2008 (Report September 2011). CE Delft, Delft, Netherlands. Church, R.L., Sorensen, P., 1994. Integrating Normative Location Models into GIS: Problems and Prospects with the p-Median Model (Technical Report 94-5). University of California, Santa Barbara, National Center for Geographic Information and Analysis. Cornillie, I., Macharis, C., 2006. Palletvervoer via de binnenvaart, een haalbaar alternatief? [Pallettransport via the inland waterways, a feasible alternative?]. In: Despontin, M., Macharis, C. (Eds.), Mobiliteit en (groot)stedenbeleid. VUBPress, pp. 207–238. Daganzo, C.F., 2005. Logistics System Analysis, fourth ed. Springer, Heidelberg, p. 296. Danielis, R., Marcucci, E., Rotaris, L., 2005. Logistics managers stated preferences for freight service attributes. Transportation Research Part E 41 (3), 201–215. De Munck, L., 2010. Palletvervoer via de binnenvaart: een business met toekomst. [Pallettransport via the Inland Waterways: A Business of the Future]. Provinvial Academy Limburg, Log-IC, Hasselt. den Boer, L.C., Otten, M., van Essen, H., 2011, STREAM International Freight 2011: Comparison of Various Transport Modes on a EU Scale with the STREAM Database. CE Delft, Delft, Netherlands. Dijkstra, E.W., 1959. A note on two problems in connexion with graphs. Numerische Mathematik 1, 269–271. ESRI, 2010. ARCMAP 10. Network Analyst Users Guide. Essenciál Supply Chain Architects, 2011. Comparative Quantified Cost Analysis between Inland Navigation and Road Transport, for Palletised Goods. N554/ 2010 – Belgium. Preliminary Study for Waterwegen en Zeekanaal NV and NV De Scheepvaart. European Commission, 2011. EU Transport in Figures: Statistical Pocketbook 2011. Publications Office of the European Union, Luxembourg. Freight Best Practice, 2005. Key Performance Indicators for the Pallet Sector: Benchmarking Guide. Department for Transport, London. Groothedde, B., Ruijgrok, C., Tavasszy, L., 2005. Towards collaborative, intermodal hub networks: a case study in the fast moving consumer goods market. Transport Research Part E 41, 567–583. Hillsman, E.L., 1984. The p-median structure as a unified linear model for location– allocation analysis. Environment and Planning A 16 (3), 305–318. Kayikci, Y., 2010. A conceptual model for intermodal freight logistics centre location decisions. Procedia Social and Behavioral Sciences 2, 6297–6311.
53
Kreutzberger, E., Macharis, C., Woxenius, J., 2006. Intermodal versus unimodal road freight transport – a review of comparisons of the external costs. In: Jourquin, B., Rietveld, P., Westin, L. (Eds.), Towards Better Performing Transport Systems. Taylor and Francis, London, pp. 17–42. LOGIQ, 2000. Final Report for Publication of LOGIQ Research Project (Contract No.: IN-97-SC.2237). Gruppo CLAS, Milan. Mommens, K., Macharis, C., 2012. Pallets on the Inland Waterways – Policy Paper (Legal Depot Number D/2012/11.528/13). Antwerp: Steunpunt Goederen- en Personenvervoer. Pekin, E., Macharis, C., Meers, D., Rietveld, P., 2012. Location analysis model for Belgian intermodal terminals: importance of the value of time in the intermodal transport chain. Computers in Industry 64, 113–120 (special issue). Poppink, P., 2005. Project roept vragen op: Distrivaart laat beperkingen modal shift zien [Project evokes some questions: Distrivaart shows the limits of the modal shift]. Transport en Logistiek 4, 26–27. Ricci, A., Black, I., 2005. The social costs of intermodal freight transport. Transport Economics 14, 245–285. Rutten, B.J.C.M., 1998. The design of a terminal network for intermodal transport. Transport Logistics 1 (4), 279–298. Service d’études techniques des routes et autoroutes (Sétra), 2008. Le transport fluvial de palettes à Paris: une activité peu ordinaire [Barge transport of pallets in Paris: a rare activity]. Revue Transports. May 2008 (Nr. 5). Ministère de l’écologie, du développement et de l’aménagement durables, Paris. Teitz, M.B., Bart, P., 1968. Heuristic methods for estimating the generalized vertex median of a weighted graph. Operations Research 16, 955–961. Van Dorsser, J.C.M., 2004. Economische Haalbaarheid van Palletvaart [Economic Feasibility of Pallettransport via the Inland Waterways]. Master Thesis, Erasmus University, Rotterdam. Van Lier, T., Macharis, C., 2013. Emissions of barge transport: a life-cycle assessment approach. In: Proceedings of 13th World Conference on Transport Research. Rio de Janeiro, Brazil, 15–18 July 2013 (to appear in selected proceedings). Vannieuwenhuyse, B., Gelders, L., Pintelon, L., 2003. An online decision support system for transportation mode choice. Logistics Information Management 16 (2), 125–133. Verbeke, F., Cornillie, I., Macharis, C., 2007. De derde golf in de binnenvaart: palletvervoer succesvol getest in de praktijk [The third wave in barge-transport: pallettransport successfully tested in practice]. In: Vervoerslogistieke Werkdagen 2007. Nautilus Academic Books, Grobbendonk, Belgium, pp. 376– 389. Verbeke, F., Macharis, C., Mommens, K., 2012. Shifting pallets to the inland waterways: the combination of theory and practice. In: Weijers, S., Dullaert, W. (Eds.), Bijdragen Vervoerslogistieke werkdagen, 29–30 November. Zelzate University Press, Venlo, pp. 331–344. Vlaams Instituur voor Mobiliteit (VIM), 2012a. De voornaamste resultaten van de proefvaarten van ‘‘Build Over Water’’ [The Main Results of the Test Bargetransports of the ‘‘Build Over Water’’]. VIM, Diepenbeek. Vlaams Instituut voor Mobiliteit (VIM), 2012b. Eindrapport Modelproject en Distributieanalyse Build over Water [Final Report of the Project and Distribution-analysis Build Over Water] (Project Nr. P029). VIM, Diepenbeek. Vrije Universiteit Brussel (VUB), COMiSOL, 2006. Haalbaarheidsstudie voor de concrete implementatie van de binnenvaart voor vervoer van pallets en de daarmee verbonden stadsbevoorrading [Feasibility-analysis for the Concrete Implementation of Pallettransport via the Inland Waterways and the Related City Distribution]. Brussels.