This white paper presents a spatial decision support system (SDSS) aimed at generating optimized vehicles routes for multiple vehicles routing problems that involves serving the demand located at nodes of a transportation network. The SDSS incorporates MapPointTM (cartography and network data), a database and a metaheuristic developed generate routes.

1. SELL - Smart Energy for
Leveraging LPG use
White Paper
Acknowledgements
This work has been framed under the ISA Academy, and totally supported by the SELL - Smart Energy for
Leveraging LPG use project (QREN).
Keywords:
Vehicle routings, decision support systems, heuristic, optimization routings.
Abstract:
This paper presents a spatial decision support system (SDSS) aimed at generating optimized vehicles routes for multiple
vehicles routing problems that involves serving the demand located at nodes of a transportation network. The SDSS
incorporates MapPointTM (cartography and network data), a database and a metaheuristic developed to generate routes.

2. Smart Energy for Leveraging LPG use
Abstract
This white paper presents a spatial decision support
system (SDSS) aimed at generating optimized
vehicles
routes
for
multiple
vehicles
routing
problems that involves serving the demand located at
nodes of a transportation network. The SDSS
incorporates MapPointTM (cartography and network
data), a database and a metaheuristic developed
generate routes.
Keywords:
Vehicle routings, decision support
systems, heuristic, optimization routings.
ii

3. Smart Energy for Leveraging LPG use
1 Introduction
Transportation of goods and services imposes considerable cost on both the public and private
sector for the economy as well as he environment. More efficient vehicle routing can improve
a firm’s competitive advantage, increase the efficiency of supplying public services, and
reduce energy consumption, traffic congestion and air pollution, which are growing problems
in many urban areas. Vehicle travel increased substantially in recent decades and 16 to 50% of
all air pollutants resulting from transportation [1].
This paper will present an optimization of distributed logistics of LPG - Smart Logistics, that
consist in a computing platform that allows for optimized logistics planning deliveries, for
several days, affecting vehicles to LPG tanks to supply daily, to determine efficient routes
and display them on maps. This platform adapts to each consumer’s according to specific
costumers’ needs and taking into account the need to reduce logistics costs. In this way, the
process consists of several steps: audit (survey procedures used and the specific problems);
adaptation of computer platform to the specific problem to be addressed; implementation of
a prototype in a specific area and expanding the process to a broader area, and monitoring
(evaluation of the procedures adopted and gains achieved).
Smart Logistics solution is composed by:
- Optimization algorithms of supply routes able to impressively reduce the vehicle kilometers
traveled, and consequently reduce the logistic costs, in the distribution of LPG;
- Computer platform that integrates algorithms and the database that stores the daily data
from the telemetry applied in LPG tanks. With this platform the distributors can plan
deliveries for the entire week by selecting, for each day the fuel tanks to supply and the
vehicles to affect;
- Component to visualize the routes selected.
Regarding the computer platform, this contains extremely efficient algorithms for
optimization in transport logistics, as well as layouts that allow users to make planning for
several days, including some variables. Thus, it is possible to visualize the expected levels in
each tank over several consecutive days, selecting those that intend to refuel every day, also
select the vehicles to be allocated for distribution in each day and assess the costs estimated
for each day of supply. This new approach is an innovation in this market niche.
The computer platform is also simple to use, easy to parameterization (containing only the
parameters strictly necessary to treat the problem in question) and low cost service.
The SmartLogistic consists of an integrated platform and adjusted to each costumer in order
to reduce logistics costs in the distribution of LPG. Thus it will be possible to avoid the use of
multiple platforms by the customer (route management software, database for client’s
management and information system telemetry), from different vendors, which sometimes
present difficulties interconnection / communication.
1

4. Smart Energy for Leveraging LPG use
2 State-of-the art
Urban freight transportation is on one hand an important economic activity but on the other
hand is rather disturbing /traffic congestion, noise and other environmental impacts). Issues
related to freight transportation are pertinent in an urban context (where the number of
vehicles, congestion and pollution levels are increasing fast) and therefore they need to be
well understood and quantified.
In what concerns environmental impacts, some recent studies emphasize the optimization of
route choice based on the lowest total fuel consumption and thus the emission of CO2 [7].
Many see the need for a threefold strategic approach: improving fuel economy, decreasing
vehicle miles of travel (VMT) and lowering the carbon content of fuels.
Woodcock et al. [8] also refer to several main strategies jointly required for moving to lowcarbon transport, being shortening trip distance one of them.
The importance of transportation problem has been acknowledged by the scientific
community. However, the optimization of transportation routing is computationally
intractable for most real-world problems (e.g. [9, 10])
It was estimated that in Western Europe the costs associated with transport have
represented, in 2000, about 7.3% of its GDP, and the road transport sub-sector accounts for
about 84% of this value [2]. Some of these costs are supported directly by operators (costs
associated with fleet vehicles, crews, etc.) and other convert into indirect costs that are
passed on society and on environment (costs associated with greenhouse gas emissions,
consumption of energy resources not renewable, traffic congestion, accidents, etc.). Efficient
management of fleets of vehicles perform an important role in reducing these costs and
contributes, in particular, to achieve the objective set out in the strategy of 20-20-20.
The problems involving the optimization of vehicle routes are the most studied in Operations
Research. The basic mathematical problems are the famous "Traveling Salesman Problem" and
"Chines Postman Problem". New versions of these two problems have been proposed over the
recent years (notably, the "Vehicle Routing Problem" and "Arc Routing Problem") in order to
make them more approximate to reality problem to deal with.
The mathematical models that currently exist to determine the optimum solution to such
problems have a high computational complexity, implying generally unacceptable execution
times (which may require days). In this way, have been developed heuristic algorithms that
obtain approximate solutions in acceptable computational times (in the order of minutes).
Some of these algorithms are based on behavior observed in nature, such as genetic
algorithms [3] and those based on optimization with ant colonies [4,5, 15], and others are
based on strategies supported almost exclusively in determination of shorter journeys [6].
This has justified the development of heuristic approaches to generate (optimal or near
optimal) solution in acceptable computer times. As a consequence, the design and
implementation of exact and heuristic algorithm for such problem constitute an interesting
challenge for operations research (OR) and transportation science, both from methodological
perspective and practical decision support purposes to address all the issues mentioned
above. The potential benefits of OR models and methods applied to transportation systems,
having in mind the implementation in practice, has constitute a research avenue followed by
2

5. Smart Energy for Leveraging LPG use
the authors, namely concerning the development of strategies to deal with real-world urban
vehicles collection/delivery problems [11,12] and new approaches for routing problems [13,
14, 15]
Due to the data requirements and the complexity of urban planning and transportation
problems, there has been a growing interest in the use of decision support system (DSS) to
analyze them at the operational (e.g., [16,17]), the tactical (e.g., [18]) and the strategic
planning levels (e.g., [19,20]). Adequate graphical interfaces are important to represent
solutions in routing problems given their strong spatial component. Information and
communication technologies (ICT) can play an important role for constructing tools
embedding algorithms, graphical interfaces and access to remote data through the Internet.
Due to the spatial nature of these problems, geographical information system (GIS) have been
a natural components of such DSS as they are important tools for collecting, organizing, and
displaying spatial data in a layer variety of planning applications, such as in vehicle touting
problems [21, 22].
Hans [23] enhanced the importance of the development of GIS for urban transportation
planning and modeling, including network based urban transportation planning and the
incorporation of network data into a GIS framework in order to have a high-speed interactive
system suitable for providing near real-time alternatives and policy analysis. Although
transportation research has been “late to embrace GIS as a key technology to support its
research and operational needs” [24], there has been an increase of such research in recent
years. Much of this research also incorporates exact and heuristic solution algorithms with the
GIS (Geographical information system)
The development of decision support tools profiting from state-of the-art ICT is an important
avenue of research. World Wide Web technologies have transformed the design,
development, implementation and deployment of DSS; however, it is recognized that the use
of Web-based computation to deploy DSS applications for remote access remains less common
[25]. In the field of transportation, some recent developments can be found, e.g., Ray [26]
has developed a web based spatial DSS for managing the movement of oversize and
overweight vehicles over highways.
We present a -SDSS integrating optimization methodologies for the LPG distribution The
system can be used for short-term analysis (e.g., the design of daily vehicle routes) and longterm analysis (e.g., deciding how many vehicles to operate in a fleet).
The most complete solutions known to the area of distribution management fuel integrate
different platforms (software route management, database management to costumers and
computer system telemetry), from different vendors, which sometimes present difficulties
interconnection / communication. On the other hand, usually computer applications
management routes used are too costly, too broad, difficult parameterization and not always
cover all the specifics of costumers.
This paper aims to shown as innovative platform developed by ISA – Smart Logistics, that
integrated all of these components in one, and adaptive to each costumer in order to reduce
logistics costs in the distribution of LPG.
3

6. Smart Energy for Leveraging LPG use
3 Solution Design
The SDSS is a standalone application which integrates a module containing the algorithms, the
MapPointTM component (that allows users to view, edit and integrate maps) and a local
database (Figure 1).
Every day, the tanks with the ISA telemetry equipment installed send their levels to a server
in the company. After that, the SDSS request these information, via a web service, and store
it in the local database. The data from the other tanks (without telemetry equipment) are
imported to the local database via excel files. These data requires user validation before be
imported. Users can generate several planning for each day and then select one of them to be
executed by the vehicle’s crew, based on the routes information.
The SDSS contains 3 main modules: import, planning and reports. Import module is used to:
a) Import tanks data from excel files;
b) Import data from the executed routes (e. g., route length, delivery schedule and delivery
quantities), used to compare planned routes and executed routes.
In the planning module users can perform the following tasks:
a) Select the tanks that need to be serviced in each day (Figure 2);
b) Introduce the required amounts of gas on each select tank (Figure 2);
c) Select the vehicles that can be used by the algorithm (solutions generated by the algorithm
can discard some vehicles selected by the users if not necessary);
d) See the routes generated by the algorithm on the maps and some related information
(route start date, route length, route duration, percentage of vehicle load, km/ton and
€/ton) (Figure 2);
e) Manually update the generated routes based on its configuration displayed on the map (e.
g., swap serviced tanks between routes, add a new tank to be serviced to an existing route,
remove a tank serviced on an existing route and add a new available vehicle to the solution).
In the reports module users can create:
a) Reports with information for each planned route contained in a selected planning (e. g.,
delivery sequence, tanks coordinates and planned quantities);
b) Reports comparing the planned routes and the executed routes (e. g., route length, route
duration, delivered quantities, number of deliveries, total delivery time, km/number of
deliveries and €/ton).
4

7. Smart Energy for Leveraging LPG use
3.1 Initial solution
The heuristic used to obtain the initial solution is based on the path-scanning heuristic [28].
However, the original version had to be adapted to deal with service located at nodes,
instead of arcs, and with additional real-world constraints, such as:
a) The route “drop-off point” is not at the same location as the depot where the vehicles
start and end their shifts;
b) Vehicles can serve more than one route in a day (all routes for a particular vehicle must
include a “drop-off point” and only the last route for each vehicle must return to the starting
depot immediately after visiting the “drop-off point);
c) Vehicles can begin one route in one day and finish it in the day after;
d) The vehicle’s crew has limit shift duration by day, which represents the maximum hours
that they can work that day;
f) Some tanks cannot be serviced by all types of vehicles – the greatest vehicles cannot
circulate in the tiniest locals.
The greatest vehicles selected by the user are the first that the algorithm tries to assign to
the solution, because they have the best transportation ratio cost/capacity.
Similar to the original heuristic, each route is obtained by adding one node at a time, instead
of an arc, according to a given criteria. In this case, to select the first tank to be serviced in
each route users can select one of the following criteria:
a) The one requiring the greatest amount of gas;
b) The farthest from the vehicle depart local.
To select the next tanks to be serviced in the routes users can select one of the following
criteria:
a) The closest to the last tank added to the route;
b) Random selecting one of the closest tanks from the last tank added to the route.
3.2 Final optimization
A final optimization step is available in the SDSS. The goal in this step is to improve the initial
solution (i. e., reducing the total travelled distance), by evaluating “neighborhood moves”.
All the constraints referred in the previous section must be satisfied also in the final solution.
Three types of moves were considered:
a) Removal and insertion of a tank – Evaluate the removal of each tank from the current route
and its insertion in another route or in another position in the current route;
5

8. Smart Energy for Leveraging LPG use
b) Swap tanks – Evaluate the swap of two tanks serviced in different routes or in the same
route;
c) Removal and insertion of an entire route – Evaluate the removal of each route from the
current vehicle and its insertion in another available vehicle.
In each iteration of the final optimization if the evaluated move improves the current solution
then it is updated and all the moves are evaluated again.
The stop criterion of this step is the maximum CPU time defined by the user in the SDSS or
the evaluation of all moves without any improvement.
6

9. Smart Energy for Leveraging LPG use
4 References
[1] - L. Dablanc, Goods transport in large European cities: Difficult to organize, difficult
to modernize, Transportation Research Part A 41 (3) (2007) 280–285.
[2] - Federação Europeia de Transportes e Ambiente, 2005, "The Eurovignette Directive.
Background
briefing".
Disponível
em:
www.transportenvironment.org/Publications/prep_hand_out/lid:359
[3] - Lacomme, P., Prins, C., Ramdane-Chérif, W., 2004a, "Competitive memetic algorithms
for arc routing problems", Annals of Operations Research, Vol. 131, pp. 159-185.
[4] - Lacomme, P., Prins, C., Tanguy, A., 2004b, "First competitive ant colony scheme for the
CARP",
Research
Report
LIMOS/RR-04-21.
Disponível
em:
http://www.isima.fr/limos/publi/RR-04-21.pdf
[5] - Reimann, M., Ulrich, H., 2006, "Comparing backhauling strategies in vehicle routing using
ant colony optimization", Central European Journal of Operations Research, Vol. 14, No. 2,
pp. 105-123.
[6] - Santos, L., Coutinho-Rodrigues, J., Current, J.R., 2009, "An improved heuristic for the
capacitated arc routing problem", Computers and Operations Research, Vol. 36, No. 9, pp.
2632-2637.
[7] - E. Ericsson, H. Larsson, K. Brundell-Freij, Optimizing route choice for lowest fuel
consumption: Potential effects of a new driver support tool, Transportation Research Part C
14 (6) (2006) 369–383.
[8] - J. Woodcock, D. Banister, P. Edwards, A.M. Prentice, I. Roberts, Energy and transport,
Lancet 370 (9592) (2007) 1078–1088.
[9] - M.R. Garey, D.S. Johnson, Computers and intractability: A guide to the theory of NPcompleteness, W.H. Freeman, New York, 1979.
[10] T.L. Magnanti, R.T. Wong, Network design and transportation planning: Models and
algorithms, Transportation Science 18 (1) (1984) 1–55.
[11] - J. Coutinho-Rodrigues, N. Rodrigues, J. Clímaco, Solving an urban routing problem
using heuristics: A successful case study, International Journal of Computer Applications in
Technology 6 (2–3) (1993) 176–180.
[12] - L. Santos, J. Coutinho-Rodrigues, J.R. Current, Implementing a multi-vehicle multi
route spatial decision support system for efficient trash collection in Portugal,
Transportation Research Part A 42 (6) (2008) 922–934.
[13] – L. Santos, J. Coutinho-Rodrigues, J.R. Current, An improved solution algorithm for the
constrained shortest path problem, Transportation Research Part B 41 (7) (2007) 756–771.
[14] – L. Santos, J. Coutinho-Rodrigues, J.R. Current, An improved heuristic for the
capacitated arc routing problem, Computers and Operations Research 36 (9) (2009) 2632–
2637.
7

10. Smart Energy for Leveraging LPG use
[15] – L. Santos, J. Coutinho-Rodrigues, J.R. Current, An improved ant colony optimization
based algorithm for the capacitated arc routing problem, Transportation Research Part B 44
(2010) 246–266.
[16] – J. Maria, J. Coutinho-Rodrigues, J. Current, Interactive destination marketing system
for small- and medium-sized tourism destinations, Tourism: An Interdisciplinary Journal 53
(2005) 45–54.
[17] – A. Simão, J. Coutinho-Rodrigues, J. Current, A management information system for
urban water supply networks, ASCE Journal of Infrastructure Systems 10 (4) (2004) 176–180.
[18] – P.A.L. Matos, P.L. Powell, Decision support for flight re-routing in Europe, Decision
Support Systems 34 (4) (2002) 397–412.
[19] – J. Coutinho-Rodrigues, J. Current, J. Climaco, S. Ratick, An interactive spatial decision
support system for multiobjective HAZMAT location-routing problems, Transportation
Research Record 1602 (1997) 101–109.
[20] - F. Űlengin, Ş. Őnsel, Y. Topçu, E. Aktaş, K. Őzgur, An integrated transportation decision
support system for transportation policy decisions: The case of Turkey,
Transportation Research Part A 41 (1) (2007) 80–97.
[21] - Z.R. Peng, R. Huang, Design and development of interactive trip planning for webbased
transit information systems, Transportation Research Part C 8 (1–6) (2000) 409–425.
[22] - L. Santos, J. Coutinho-Rodrigues, J.R. Current, Implementing a multi-vehicle multiroute
spatial decision support system for efficient trash collection in Portugal,
Transportation Research Part A 42 (6) (2008) 922–934.
[23] – Z. Hans, R. Souleyrette, GIS and network models: Issues for three potential
applications, Journal of Advanced Transportation 29 (3) (1995) 355–373.
[24] – J.C. Thill, Geographic information systems for transportation in perspective,
Transportation Research Part C 8 (1–6) (2000) 3–12.
[25] – L. Santos, J. Coutinho-Rodrigues, J.R. Current, An improved ant colony optimization
based algorithm for the capacitated arc routing problem, Transportation
Research Part B 44 (2010) 246–266.
[26] – J.J. Ray, A web-based spatial decision support system optimizes routes for
oversize/overweight vehicles in Delaware, Decision Support Systems 43 (4) (2007) 1171–1185.
[27] - J. Coutinho-Rodrigues, N. Rodrigues, J. Clímaco, Solving an urban routing problem using
heuristics: A successful case study, International Journal of Computer Applications in
Technology 6 (2–3) (1993) 176–180.
[28] – B.L. Golden, J. DeArmon, E.K. Baker, Computational experiments with algorithms for a
class of routing problems, Computers and Operations Research (10) (1983) 47–59.
8