Available via license: CC BY 4.0
Content may be subject to copyright.
Proceedings of the XIII International
Workshop on Locational Analysis
and Related Problems
Granada, Spain
4-6 September 2024
arXiv:2409.06397v1 [math.OC] 10 Sep 2024
Proceedings of the XIII International
Workshop on Locational Analysis
and Related Problems
Granada, Spain
4-6 September 2024
PROCEEDINGS OF THE XIII INTERNATIONAL
WORKSHOP ON LOCATIONAL ANALYSIS
AND RELATED PROBLEMS (2024)
Edited by
Marta Baldomero-Naranjo
Ricardo Gázquez
Miguel Martínez-Antón
Luisa I. Martínez-Merino
Juan M. Muñoz-Ocaña
Francisco Temprano
Alberto Torrejón
Carlos Valverde
Nicolás Zerega
ISBN: 978-84-09-63233-6
XIII International Workshop on Locational Analysis and Related Problems
Contents
Welcome 9
IWOLOCA 11
Committees 13
Program 15
Wednesday, September 4, 2024 15
Thursday, September 5, 2024 17
Friday, September 6, 2024 19
Invited Speakers 21
Abstracts 29
Useful Information 91
Venues 91
Social Activities 92
Partner Institutions and Sponsors 95
7
Welcome
Dear Locators,
Welcome to the XIII International Workshop on Locational Analysis and Related Problems
(IWOLOCA 2024) in Granada, a meeting organized by the Spanish Network on Locational
Analysis (REDLOCA) and the Location Group GELOCA (Spanish Society of Statistics and
Operations Research–SEIO). Our workshops aim to bring together researchers and practitioners
working in the wide area of Location Science through the proposal of models and methods
to solve theoretical and practical problems. These meetings traditionally combine a strong
scientific value with a very enjoyable and friendly come together as an enthusiastic and
energetic group, where old friends meet again and new members are heartily welcomed.
The success of these meetings over the years proves the continuing research community
interested in locational analysis and related problems. This 13th edition of the workshop
features 26 contributions covering a broad range of topics, from discrete location, over hub
location and routing, to network design and several interesting applications, including those
in cross-docking designs. During the three days of the meeting the works will be presented
in 7 sessions, with around 40 participants. The workshop will feature the participation of
three highly recognized plenary speakers who accepted our invitation to come to Granada
and share their insights with us. It is a pleasure to have them in Granada, thank you! Prof.
Marín (Universidad de Murcia) will present his work on an interesting application of location
problems to determining original DNA sequences from different fragments. Prof. Luedtke
(UW-Madison) will bring models and solution methodologies for solving a location problem
that arises when designing power generation plants based on the impact of extreme weather
events. Eventually, Prof. Salman (Koç University) will share different studies on disaster
management and locational decisions with us.
We would like to thank all the members of the Organizing Committee for their valuable help
in the organization, and we are very grateful to the colleagues of the Scientific Committee.
They acknowledge the editors of these proceedings, for the hard work committed to collecting
all the information about the workshop and giving it this fantastic shape. We would also
like to thank our graphical designer, María Martín Hersog, for her help in designing all the
amazing drawings identifying the workshop in this edition. Finally, we would like to thank
the support of the Spanish Agency of Research (AEI) through grants RED2022-134149-T
9
Granada, 4-6 September 2024 IWOLOCA 2024
funded by MICIU/AEI /10.13039/501100011033 and IMAG-Maria de Maeztu grant CEX2020-
001105-M /AEI /10.13039/501100011033, Universidad de Granada, GELOCA (SEIO), and
Institute of Mathematics (IMAG) as well.
Our workshop represents a great opportunity to build new and enrich existing relationships
and to share experiences among locators. We wish you all a successful and fruitful meeting,
with new ideas and collaborations, and a pleasant stay in Granada.
Víctor Blanco
IWOLOCA 2024 Organizing Committee, chair
10
IWOLOCA
The XIII International Workshop on Locational Analysis and Related Problems will take place
during September 4–6, 2024 in Granada (Spain). It is organized by the Spanish Location
Network, REDLOCA, and the Location Group, GELOCA, from the Spanish Society of
Statistics and Operations Research (SEIO). The Spanish Location Network is a group of 100+
researchers from several Spanish universities organized into 7 thematic groups. The Network
has been funded by the Spanish Government since 2003.
The topics of interest are location analysis and related problems. This includes location models,
networks, transportation, logistics, exact and heuristic solution methods, and computational
geometry, among many others.
Previous meetings
One of the main activities of the Network is a yearly meeting aimed at promoting communica-
tion among its members and between them and other researchers and contributing to the
development of the location field and related problems. The last meetings have taken place in
Edinburgh (September 7–8, 2023), Elche (January 31–February 1, 2022), Sevilla (January
23–24, 2020), Cádiz (January 20–February 1, 2019), Segovia (September 27–29, 2017),
Málaga (September 14–16, 2016), Barcelona (November 25–28, 2015), Sevilla (October 1–3,
2014), Torremolinos (Málaga, June 19–21, 2013), Granada (May 10–12, 2012), Las Palmas
de Gran Canaria (February 2–5, 2011) and Sevilla (February 1–3, 2010).
11
Committees
Scientific Committee
Mari Albareda Sambola Universidad Politécnica de Cataluña
Víctor Blanco Universidad de Granada
David Canca Universidad de Sevilla
Sergio García Quiles University of Edinburgh
Jörg Kalcsics University of Edinburgh
Mercedes Landete Universidad Miguel Hernández de Elche
Teresa Ortuño Universidad Complutense de Madrid
Blas Pelegrín Universidad de Murcia
Justo Puerto Albandoz Universidad de Sevilla
Antonio M. Rodríguez-Chía Universidad de Cádiz
Juan José Salazar Universidad de La Laguna
Organizing Committee
Víctor Blanco Universidad de Granada
Ricardo Gázquez Universidad Carlos III de Madrid
Gabriel González Universidad de Granada
Miguel Martínez-Antón Universidad de Granada
Román Salmerón Universidad de Granada
13
Program
I: Invited Talk, C: Contributed Talk
Wednesday, September 4, 2024
4:20–4.30 Registration
4:30–4:45 Opening session
4:45–5:45 Plenary talk I
IAlfredo Marín
U. Murcia, Spain
A new application of discrete
location: DNA sequence assembly
5:45–6:00 Coffee
6:00–7:40 Session 1: Applications
6:00–6:25 C Alberto Japón
U. Sevilla, Spain
Classification forests via
mathematical programming: case
study in obesity detection
6:25–6:50 C
Ramón
Piedra-de-la-Cuadra
U. Sevilla, Spain
Optimization approaches to
scheduling working shifts for train
dispatchers
6:50–7:15 C Eduardo Pipicelli
U. Naples Federico II, Italy
A mathematical model for designing
a postal banking network
7:15–7:40 C David Canca (chair)
U. Sevilla, Spain
A bi-level metaheuristic for the
design of hierarchical transit
networks in a multimodal context
8:30 Welcome Cocktail: BHeaven Granada
15
XIII International Workshop on Locational Analysis and Related Problems
Thursday, September 5, 2024
9:00–10:00 Plenary talk II
I
Jim Luedtke
U. Wisconsin-Madison,
USA
Generator Expansion to Improve
Power Grid Resiliency and Efficiency:
A Case Study in Location Analysis
Under Uncertainty
10:00–11:15 Session 2: Covering
10:00–10:25 C
Marta Baldomero Naranjo
U. Cádiz, Spain
The maximal covering location
problem with edge downgrades
10:25–10:50 C Ricardo Gázquez
U. Carlos III Madrid, Spain
An upgrading approach for the
multi-type Maximal Covering
Location Problem
10:50–11:15 C
Antonio Rodríguez Chía
(chair)
U. Cádiz, Spain
New results in the covering tour
problem with edge upgrades
11:15–11:35 Coffee
11:35–11:50 Groups meetings
11:50–1:30 Session 3: Discrete Location
11:50–12:15 C Concepción Domínguez
U. Murcia, Spain
Stable Solutions for the Capacitated
Simple Plant Location Problem with
Order
12:15–12:40 C Sophia Wrede
RWT Aachen U., Germany,
Cover-based inequalities for the
single-source capacitated facility
location problem with customer
preferences
12:40–1:05 C
Juan José Salazar
González
U. La Laguna, Spain
Solving a multi-source capacitated
facility location problem with a
fairness objective
1:05–1:30 C
José Fernando Camacho
Vallejo (chair)
Tec Monterrey, Mexico
A Reformulation for the Medianoid
Problem with Multipurpose
Trips
17
Granada, 4-6 September 2024 IWOLOCA 2024
1:30–3:15 Lunch
3:15–4:30 Session 4: Network I
3:15–3:40 C Francisco Temprano
U. Sevilla, Spain
New advances in hypergraph
structure analysis
3:40–4:05 C Gabriel González
U. Granada, Spain
Expanding Reaction Networks
through Autocatalytic Subnetworks
that Maximize Growth Factor
4:05–4:30 C Carlos Valverde (chair)
U. Sevilla, Spain
The Hampered K-Median Problem
with Neighbourhoods
4:30–5:45 Session 5: Hub Location and Routing
4:30–4:55 C
Juan Manuel
Muñoz-Ocaña
U. Cádiz, Spain
Formulations and resolution
procedures for upgrading hub
networks
4:55–5:20 C Inmaculada Espejo
U. Cádiz, Spain
A further research on Stochastic
Single-Allocation Hub Location
Problem
5:20–5:45 C
Mari Albareda (chair)
U. Politécnica Catalunya,
Spain
A single vehicle location routing
problem with pickup and delivery
5:45 Group Picture & Guided Tour
18
XIII International Workshop on Locational Analysis and Related Problems
Friday, September 6, 2024
9:00–10:00 Plenary talk III
ISibel Salman
Koç U., Turkey
Location, Allocation, Routing and
Network Design in Humanitarian
Operations
10:00–11:15 Session 6: Routing
10:00-10:25 C
Inigo Martin Melero
U. Miguel Hernández Elche,
Spain
The Effect of Budget Limiting on
the Linear Ordering Problem
10:25–10:50 C Aitor López-Sánchez
U. Rey Juan Carlos, Spain
Synchronization routing for
agricultural vehicles and implements
10:50–11:15 C Paula Segura (chair)
U. Valencia, Spain
Exact approaches for the Chinese
Postman Problem with
load-dependent costs
11:15–11:35 Coffee
11:35–12:15 Groups meetings
12:15–1:30 Session 7: Combinatorial Optimization
12:15–12:40 C Alberto Torrejón
U. Sevilla, Spain
Modeling ordered interactions in
Location Problems
12:40–1:05 C Miguel Martínez-Antón
U. Granada, Spain Finding a smallest mediated set
1:05–1:30 C
Martine Labbé (chair)
U. Libre de Bruxelles,
Belgium
Novel Valid Inequalities for
Chance-Constrained Problems with
Finite Support
19
Granada, 4-6 September 2024 IWOLOCA 2024
1:30–3:15 Lunch
3:15–4:30 Session 8: Cross-docking door design
3:15–3:40 C Aitziber Unzueta
U. Pais Vasco, Spain
Cross-docking platforms design and
management under uncertainty
3:40-4:05 C Laureano Escudero
U. Rey Juan Carlos, Spain
Cross-docking platforms design and
distributionally robust optimization
4:05-4:30 C
Maria Araceli Garin
(chair)
U. Pais Vasco, Spain
Cross-docking platforms design and
mixed binary quadratic model for
distributionally robust optimization
4:30–4:45 Closing Session
4:45–5:15 Coffe Break
5:15–5:45 REDLOCA Meeting: Spanish Network on Locational Analysis
9:00 Gala Dinner: Carmen de la Victoria
20
Invited Speakers
XIII International Workshop on Locational Analysis and Related Problems
A new application of discrete location: DNA sequence assembly
Alfredo Marín
Universidad de Murcia, amarin@um.es.
Keywords: Facility location, Discrete Location
Given fragments of several copies of a unique DNA sequence, "sequence assembly" is the
problem of determining the original sequence from these fragments. Integer Programming
formulations coming from the vehicle routing field have been used sometimes to solve the
problem. In this talk we try to re-interpret the problem as a kind of discrete location problem
and design formulations based on this new interpretation.
23
XIII International Workshop on Locational Analysis and Related Problems
Generator Expansion to Improve Power Grid Resiliency and Effi-
ciency: A Case Study in Location Analysis Under Uncertainty
James Luedtke
a,*
, Ramsey Rossmann
a
, Mihai Anitescu
b
, Julie Bessac
b
, Mitchell
Krock b, Line Roald a
aUniversity of Wiconsin-Madison, jim.luedtke@wisc.edu
bArgonne National Laboratory,
*Presenting author.
Keywords: Facility Location, Power Grid, Resiliency, Stochastic Programming
Introduction
We consider the problem of choosing power generator locations in a power grid to yield
a system that is both efficient on average and resilient to extreme weather events. This
problem is challenging due to the combination of discrete decisions and the need to consider
performance of the system under rare high-impact outcomes. We use this problem as a
case study to illustrate modeling principles and solution techniques that may be useful more
broadly in location analysis problems under uncertainty. Specifically, we introduce a stochastic
integer programming modeling framework, advocate for a bi-objective modeling approach for
balancing efficiency and resiliency, introduce a conditional sampling method for addressing
the challenge of low-probability high-impact events, discuss methods for solving the resulting
approximation models, and illustrate the importance of modeling spatial dependence of the
uncertain parameters in the system.
Sample Results
Figure 1 displays the results of our proposed model and compares it to alternatives. This figure
displays the two relevant metrics of interest, average cost and the average load shed in the
0.01% highest load-shed hours, which is a measure of the resiliency of the system to extreme
events. Varying model parametrs allows each method to obtain solutions that trade-off these
two objectives. Model ‘base’ is a traditional two-stage stohastic programming model that
minimizes expected cost, where the cost includes a penalty on load shed. Model ‘BO-CVaR’
is a bi-objective model that explicitly considers the two objectives of minimum expected
cost and minimum load shed in extreme scenarios, and is solved by standard sample average
approximation (SAA). This method fails to find solutions of low risk due to the challenge in
estimating the risk measure in extreme events when using SAA. Model ‘BO-CVaR-Cond’ is the
25
Granada, 4-6 September 2024 IWOLOCA 2024
same model as ‘BO-CVaR’ but adjusts the sampling to generate scenarios from a distribution
conditional on the temperatures being in either the 1% extreme lowest or highest scenarios.
This method is much more capable of generating solutions with low risk of load shedding.
Figure 1:
Comparison of efficient frontiers obtained from traditional stochastic programming
model, bi-objective model, and bi-objective model with conditional sampling.
Figure 2 illustrates the importance of modeling spatial dependence in our models – without
doing so, the model is unable to generate solutions with low risk of load shedding.
(a) Dependent model. (b) Independent model.
Figure 2:
Comparison of efficient frontiers obtained from model that considers spatial depen-
dence and model that assumes spatial independence.
26
XIII International Workshop on Locational Analysis and Related Problems
Location, Allocation, Routing and Network Design in
Humanitarian Operations
Sibel Salman
Koç University, ssalman@ku.edu.tr.
Keywords: Humanitarian Operations, Facility Location, Network Design, Vehicle Routing
Introduction
Due to the increasing number of people affected by disasters worldwide and the increase in
weather-related disasters, developing mathematical models and solution methodology has
become increasingly important to reduce risk and ensure efficient response. We will present
two studies on disaster preparedness and response: i) relief aid distribution after a natural
disaster and ii) providing vaccination to remote areas during a pandemic.
Designing an efficient and equitable relief aid distribution network with
sorting and recycling facilities
In post-disaster response, relief items are delivered to meet immediate needs and alleviate
suffering. Alongside contributions from organizations, individuals often send in-kind donations,
which can be unsuitable and complicate management, leading to resource wastage and
delays. Establishing sorting facilities can mitigate these issues by ensuring only relevant items
reach distribution points, with others stored or recycled. We focus on designing an efficient
and equitable relief aid distribution network and investigate the impact of relief aid sorting
and reverse supply chain strategies by optimizing a network configuration with and without
sorting facilities. The nodes consist of donation centers as supply points, sorting centers as
intermediate facilities, points of distribution, relief agency warehouses, and recovery centers
as destination points. Road and air shipment options exist between these locations, with
helicopters offering faster delivery but at higher costs.
In the case with sorting centers and recycling, donated items are sent to sorting centers,
where they are categorized into three types: Type I (urgent items), Type II (useful for future
disasters), and Type III (broken or unusable items). Type I items are quickly dispatched to
local distribution points, Type II items are stored in relief agency warehouses, and Type III
items are sent to recovery centers for processing and revenue generation. We use a three-
stage stochastic programming model to optimize the location and shipment decisions. First,
sorting center locations are determined pre-disaster under demand and supply uncertainty.
27
Granada, 4-6 September 2024 IWOLOCA 2024
Post-disaster, realized demand and travel times dictate shipment quantities to sorting centers.
Once sorted, shipment quantities, destinations for each item type, and transportation modes
are determined. The goal is to minimize total expected costs, including setup, workforce,
shipment, and unsatisfied demand costs, offset by revenue from Type III items. In the unsorted
donations case, donations with unknown mixes of Type I, II, and III items are sent directly
from donation centers to PODs. At the PODs, items are categorized upon arrival: Type I
items go to victims, while Types II and III are discarded. Thus, only a portion of shipped
items meets demand. The total supply chain cost includes shipment costs and the penalty
for unsatisfied demand. We use a single-stage stochastic programming model to determine
shipment quantities from donation centers to PODs. We applied the two models to the
anticipated Istanbul earthquake scenario and compared sorted and unsorted donation cases
based on total cost and the latest shipment time.
Routing and Scheduling of Mobile Vaccination Services in a Pandemic
During a pandemic, the widespread distribution of vaccines is crucial to avoid an explosive
increase in infections. However, some people have difficulty accessing vaccination services at
fixed locations, such as those living in rural areas, disabled or elderly people who are immobile,
and refugees, for which it is observed that vaccination rates are lower. Fixed vaccination
services are extended with mobile facilities that get close to the locations of such people. We
optimize the routes and schedules of the mobile facilities over a medium-term planning horizon
to minimize the number of people that cannot be reached due to capacity restrictions, the
total lateness in vaccination, the total distance traveled by the mobile facilities, and inequity
over the locations. We formulate a multi-period selective routing model and incorporate several
alternative approaches for representing equity. We develop a heuristic solution approach to
solve the case of COVID-19 vaccination of Syrian refugee groups, in addition to Turkish
citizens living in different neighborhoods of the city of Gaziantep in Turkey. We develop a
three-stage matheuristic and a metaheuristic to solve this large-scale instance and derive
insights on service quality via various analyses.
28
Abstracts
XIII International Workshop on Locational Analysis and Related Problems
A single vehicle location routing problem with
pickup and delivery
Mari Albareda a,*, Víctor Blanco b, Yolanda Hinojosa c
a
Department d’Estadística i Investigació Operativa. Universitat Politécnica de Catalunya,
Spain, maria.albareda@upc.edu
b
Institute of Mathematics (IMAG). Dpt. Quant. Methods for Economics & Business.
Universidad de Granada, Spain, vblanco@ugr.es
c
Institute of Mathematics (IMUS). Dpt. Applied Economics I. Universidad de Sevilla, Spain,
yhinojos@us.es
*Presenting author.
Keywords: Facility Location, Network Design
Due to the urgent need to combat climate change, most of the World Organizations have
recognized the importance of transitioning to cleaner and more sustainable energy sources in
the coming years to reduce greenhouse gas emissions. It is a known fact that one of the main
strategies to achieve the proposed objective is the use of biogas as an alternative renewable
energy source to carbon-based energies. In this sense, mathematical optimization is recognized
as a fundamental tool for the design and modeling of multiple logistics, transportation and
supply chain problems and, in particular, for designing robust and efficient supply chains to
integrate bioenergy into any economy [2, 3]. The paper by [1] introduces a new logistic
problem for waste-to-biogas transformation. The authors provide a general and flexible
mathematical optimization model that allows decision makers to optimally determine the
locations of different types of plants (pretreatment, biogas, and liquefaction plants), as well
as the most efficient distribution of products (waste to biomethane) along the supply chain
involved in the logistic process. They jointly integrate all these complex stages into a mixed
integer linear programming (MILP) model that attempts to minimize the overall transportation
cost of the system assuming that a limited budget is available to install all types of plants.
In this work, we further explore the model proposed in [1] by analyzing different alternatives
to incorporate routes of vehicles in different phases of this problem. Specifically, we provide
models to decide the most efficient way to determine the routes to be followed by vehicles
passing through all farms (pick-up points) and pre-treatment plants (delivery points). The
goal is to provide a mathematical optimization based framework for choosing the delivery
points to open (within a given set of potential delivery locations) in order to minimize the
31
Granada, 4-6 September 2024 IWOLOCA 2024
overall transportation costs required to collect all the demand from the pick-up points and
deliver it to one of the open delivery points using a single vehicle. Among other alternatives,
we study the problem with the following hypotheses: a single visit to each pick-up point is
allowed or multiple visits are allowed; the resulting route is required to be a cycle (it must
start and end at the same delivery point) or not; the resulting route must be connected or not
(which would imply the use of multiple vehicles, etc.). The different alternatives pose different
mathematical challenges for both modeling and solving the resulting problems. We apply
and test the different models in the pickup and delivery problem (PDP) instances TS2004t2
and TS2004t3 used in [6] and in a real-word dataset based on the region of upper Yahara
Watershed in the state of Wisconsin [4].
Bibliography
[1]
Víctor Blanco, Yolanda Hinojosa, Victor Zavala. The Waste-to-Biomethane Logistic
Problem: A mathematical optimization approach. ACS Sustainable Chem. Eng. 12, 22,
8453–8466, 2024.
[2]
Egieya Jafaru M., cucek Lidija, Zirngast Klavdija, Isafiade Adeniyi J., Pahor Bojan, and
Kravanja Zdravko. Biogas supply chain optimization considering different multi-period
scenarios. Chemical Engineering Transactions, 70:985–990, 2018.
[3]
Ida Græsted Jensen, Marie Münster, and David Pisinger. Optimizing the supply chain
of biomass and biogas for a single plant considering mass and energy losses. European
Journal of Operational Research, 262(2):744–758, October 2017.
[4]
A. M. Sampat, A. Hicks, G. J. Ruiz-Mercado, and V. M. Zavala. Valuing economic
impact reductions of nutrient pollution from livestock waste. Resources, Conservation
and Recycling, 164:105199, 2021.
[5]
Church, Richard, and ReVelle, Charles. The maximal covering location problem. Papers
in Regional Science, 32(1), 101-118, 1974.
[6]
H. Hernández-Pérez, J.J.Salazar-González. Heuristics for the one-commodity pickup-and-
delivery traveling salesman problem. Transportation Science, 38(2):245-255, 2004.
32
XIII International Workshop on Locational Analysis and Related Problems
A mathematical model for designing a postal banking network
Silvia Baldassarre a, Giuseppe Bruno a, Manuel Cavola b, Eduardo Pipicelli a,*
a
University of Naples Federico II, Italy, Department of Industrial Engineering, Piazzale
Tecchio 80, Naples, Italy,
silvia.baldassarre@unina.it
,
giuseppe.bruno@unina.it
,
eduardo.pipicelli@unina.it
bPegaso University, manuel.cavola@pegaso.it
*Presenting author.
Keywords: Facility Location, Network design, Decision support
Banking groups worldwide are significantly closing their branches while leveraging digital
channels to deliver various banking services, from basic to complex ones. This shift has raised
concerns about financial exclusion, particularly for people hesitant or reluctant to adopt digital
channels or those living in lower-income and less densely populated areas. In this scenario,
postal operators have emerged as main competitors in providing physical and face-to-face
financial services in areas without a banking presence, leveraging their capillary network of
post offices.
In this work, we propose a mathematical model to support postal providers in making decisions
about the network design for consulting services. The objective is to maximize the demand
from customers who are financially excluded due to branch closures. The postal banking
network design concerns activating consulting services in post offices to capture market share
and planning service time availability to satisfy the demand. An available budget for activating
consulting services is considered.
The model has been applied to a real case study: the rural areas of the Campania region in
southern Italy. In these areas, bank branch closures have significantly impacted customers’
accessibility to banking services over the last few years. The results demonstrate that the
model can effectively support the consulting network design process, offering an effective
framework for making informative decisions based on the preferred postal provider’s provision
policy.
33
XIII International Workshop on Locational Analysis and Related Problems
An upgrading approach for the multi-type Maximal Covering Lo-
cation Problem
Marta Baldomero-Naranjo a, Ricardo Gázquez b,*, Antonio M.
Rodríguez-Chía a
a
Universidad de Cádiz, Spain,
marta.baldomero@uca.es
,
antonio.rodriguezchia@uca.es
bUniversidad Carlos III de Madrid, Spain, ricardo.gazquez@uc3m.es
*Presenting author.
Keywords: Covering Problems, Upgrading, Maximal Covering
Covering location problems are among core problems in Location Science. These problems
arise when facilities offer services within a specific distance or maximum time. Depending on
the context, different problems may occur, but they generally fall into two categories: set
covering and maximal covering.
This presentation centers on the Maximal Covering Location Problem (MCLP), initially
introduced by Church and ReVelle ([3]). Specifically, it explores an extension of MCLP that
includes multiple types of facilities and their upgrades.
While most existing literature addresses problems within a single setting, recent research
examines the use of multiple settings within the same problem in MCLP ([2]). Conversely,
upgrading problems are becoming increasingly prominent in recent location literature. These
extensions involve coordinating facility location decisions with improvements to related infras-
tructure. In the MCLP, authors in [1] examine scenarios where upgrading the network reduces
the distance or time from a facility to a customer.
We named the model presented in this talk as the Multi-type Maximal Covering Location
Problem with Upgrading (MTMCLP-U). It extends the MCLP by incorporating the two
described approaches above: different types of facilities (in both continuous and network
settings) and two types of upgrades (for the facilities and the network).
In this model, given a set of customers—some with access to the network and some with-
out—the objective is to place a fixed number of facilities at network nodes and another fixed
number in the continuous space to maximize covered demand. Additionally, the model includes
two types of binary upgrades (yes-or-no decisions) within a common budget. In the network
setting, upgrades reduce the lengths of the edges, while in the continuous setting, upgrades
increase the coverage radius, thereby reaching more customers.
The MTMCLP-U responds to problems in real applications such as telecommunication networks.
35
Granada, 4-6 September 2024 IWOLOCA 2024
For instance, in rural areas where access to telecommunication services cannot be carried over
the wired network, repeaters are necessary. See Figure 3 for a comparison of facility locations
with (Fig. 3a) and without (Fig. 3b) upgrades.
(a) Without upgrading. (b) Upgrading in both settings.
Figure 3:
The green triangle represents a discrete facility, located only at network nodes. The
orange triangle represents a continuous facility, which can be located anywhere in
the plane. The dashed line indicates an upgraded edge, and light yellow denotes the
upgraded coverage area. Figure 3a shows the facility locations without upgrades.
Figure 3b shows the locations with upgrades, covering more points.
Bibliography
[1]
Baldomero-Naranjo, Marta, Kalcsics, Jörg, Marín, Alfredo, and Rodríguez-Chía, Antonio
M. Upgrading edges in the maximal covering location problem. European Journal of
Operational Research, 303(1), 14–36, 2022.
[2]
Blanco, Víctor, Gázquez, Ricardo and Saldanha-da-Gama, Francisco Multi-type maximal
covering location problems: Hybridizing discrete and continuous problems. European
Journal of Operational Research, 307(3), 1040–1054, 2023.
[3]
Church, Richard, and ReVelle, Charles. The maximal covering location problem. Papers
in Regional Science, 32(1), 101–118, 1974.
36
XIII International Workshop on Locational Analysis and Related Problems
The maximal covering location problem with edge downgrades
Marta Baldomero-Naranjo a,*, Jörg Kalcsics b, Antonio M. Rodríguez-Chía a
aUniversidad de Cádiz, marta.baldomero@uca.es,antonio.rodriguezchia@uca.es
bThe University of Edinburgh, joerg.kalcsics@ed.ac.uk
*Presenting author.
Keywords: Facility Location, Bilevel optimization, Heuristic algorithms
Introduction
This talk deals with an extension of the well-known Maximal Covering Location Problem
(MCLP), see [3]. It considers that an agent (attacker) can modify the length of some
edges within a budget, increasing the distance from the clients to the facilities. Under this
assumption, the Downgrading Maximal Covering Location Problem emerges as a significant
challenge, involving the interests of both the facility location planner and potential attackers.
Note that this implies that the distances in the network are modified during the optimization
problem, thus, the distance between two nodes after the downgrades will have to be computed
within the model.
This study falls within the domain of downgrading (upgrading) network problems, in which
specific elements initially treated as fixed inputs in the classical version are now transformed
into decision variables (e.g. the length of the edges). As far as we know, it is the first study
analysing edge downgrades. The version of the problem where edge upgrades are considered
has been discussed in [1, 2].
Contributions
The problem entails the strategic location of facilities within the nodes of a network to
maximise coverage while anticipating and mitigating potential attacks aimed at reducing this
coverage. Then, it comprises an interaction between two distinct actors, each with conflicting
objectives:
•
The location planner aims to maximise the covered demand while anticipating that an
attacker will attempt to reduce coverage by increasing the length of the edges.
•
The attacker seeks to maximise the demand initially covered by the facilities but left
uncovered after the downgrade. To achieve this, they can increase the length of certain
edges within a specified budget.
37
Granada, 4-6 September 2024 IWOLOCA 2024
We propose a mixed-integer linear bilevel formulation to model the problem. Moreover, we
introduce a strategy to preprocess the data to reduce the number of variables and constraints
of the formulation. Furthermore, a matheuristic algorithm to solve the problem is developed.
For this algorithm, we present some variants (levels of intensity), which allow the user to
decide the best trade-off between the computational time employed and the quality of the
solution.
Several computational experiments are carried out to illustrate the advantages of applying
the introduced model rather than using the classical MCLP and to show the potential and
limitations of the proposed matheuristic algorithm. The solutions provided by the matheuristic
are compared with the ones obtained by the general bilevel solver developed in [4].
Bibliography
[1]
Baldomero-Naranjo, M., Kalcsics, J., Marín, A., and Rodríguez-Chía, A. M. Upgrading
edges in the maximal covering location problem. European Journal of Operational Research,
303(1):14–36, 2022.
[2]
Baldomero-Naranjo, M., Kalcsics, J., and Rodríguez-Chía, A. M. On the complexity of the
upgrading version of the maximal covering location problem. Networks, 83(4):627-641,
2024.
[3]
Church, R., and ReVelle, C. The maximal covering location problem. Papers in Regional
Science, 32(1), 101-118, 1974.
[4]
Fischetti, M., Ljubić, I., Monaci, M., and Sinnl, M. A new general-purpose algorithm for
mixed-integer bilevel linear programs. Operations Research, 65(6):1615–1637, 2017.
38
XIII International Workshop on Locational Analysis and Related Problems
New results in the covering tour problem with edge upgrades
Marta Baldomero-Naranjo
a
, Andrea Mancuso
b
, Adriano Masone
b
, Antonio M. Rodríguez-
Chía a,*
a
Departamento de Estadística e Investigación Operativa, Facultad de Ciencias, Universidad
de Cádiz, Cádiz, Spain, marta.baldomero,antonio.rodriguezchia@uca.es
b
Department of Electrical Engineering and Information Technology, University of Naples
Federico II, Naples, Italy, andrea.mancuso,adriano.masone@unina.it
*Presenting author.
Keywords: Facility Location, Covering, TSP, Upgrading problem
This talk addresses a version of the Covering Tour Problem, a covering problem within the
context of location decision problems combined with a routing problem. This version considers
that the edges can be upgraded subject to a budget constraint. Therefore, the goal is to find
a minimum length tour that cover all the nodes of a graph taking into account that some
edges can be upgraded.
Introduction
Routing problems are a class of problems focused on finding the most effective way to move
goods, people, or information from one point to another. The primary objective is often
to minimize travel time, distance, or operational costs while adhering to various constraints
and requirements. These problems are central to many industries, including logistics, supply
chain management, and transportation. Among these routing problems, one of the most
fundamental is the Traveling Salesman Problem (TSP). The TSP involves finding the shortest
possible route that allows a salesman to visit each city in a given list exactly once and return
to the origin city, see [1].
Coverage problems arise when it is necessary to determine the optimal location of one or more
facilities for the production of goods or the provision of services, with the goal of meeting
the demand for goods or services at various nodes. The main objective is to minimize the
placement costs of facilities or maximize the coverage of demand points. For each facility, a
coverage radius is defined, representing the distance within which the demand for goods or
services can be satisfied.
Combining TSP and convering problems is obtained the covering tour problem (CTP). This
problem aims to determine a minimum lengh tour over a subset of nodes that can be visited,
in such a way that the tour contains all these nodes, and every node is covered by the tour,
39
Granada, 4-6 September 2024 IWOLOCA 2024
see [3].
The goal of this talk is to analyze the covering tour problem with edge upgrading (CTPEU).
This problem is an extension of the CTP where we consider the possibility of upgrading edges.
It combines the routing aspect seen in the TSP, the covering part of the CTP, and additionally
allows for the possibility of updating edges, similar to the Maximal Covering Problem with
Edge Upgrade, see [2]. Edge upgrading consists in reducing their length (travel time), typically
within certain limits, at a given cost that is proportional to the upgrade. The total cost of all
upgrades is subject to a budget constraint. On this basis, the CTPEU involves finding a tour
of minimum length subject to covering and budget constraints.
Bibliography
[1]
Applegate, D.L., Bixby R.E., Chvatal, V. and Cook, W.J. (2007). The traveling salesman
problem: a computational study, Princeton Series in Applied Mathematics.
[2] Baldomero-Naranjo, M., Kalcsics, J., Marín, A., and Rodríguez-Chía, A. M. (2022). Up-
grading edges in the maximal covering location problem. European Journal of Operational
Research, 303(1), 14-36.
[3]
Gendreau, M., Laporte, G., and Semet, F. (1997). “The covering tour problem”. Operations
Research 45(4):568-576.
40
XIII International Workshop on Locational Analysis and Related Problems
Expanding Reaction Networks through Autocatalytic Subnetworks
that Maximize Growth Factor
Víctor Blanco a, Praful Gagrani b, Gabriel González c,*
aUniversidad de Granada, vblanco@ugr.es
bUniversity of Wisconsin-Madison praful.gagrani@gmail.com
cUniversidad de Granada, ggdominguez@ugr.es
*Presenting author.
Keywords: Chemical Reaction Networks, Origin Of Life, Autocatalytic Network, Lineal
Programming, Network Design
The dynamics of Chemical Reaction Networks (CRNs) and their implications are intricate
matters with an important role in the origin of life. This work focuses on identifying autocat-
alytic subnetworks within CRNs, crucial for explicating self-replication dynamics in biological
and prebiotic systems. Autocatalysis, where a set of species catalyze their own production, is
pivotal for the emergence of life and chemical evolution. Detecting autocatalytic species within
CRNs poses a significant computational challenge, addressed here through a mathematical
optimization-based framework. Several tools have been developed to detect and enumerate
autocatalytic subnetworks by solving certain mixed integer optimization problems [1, 2]. It is
demonstrated that Discrete Optimization is a powerful tool in the search for these complex
structures. The proposed model intends to determine the most prominent autocatalytic
subnetwork within a CRN and simulate its expansion over time. Through sequential incorpora-
tion of species and reactions, the model constructs increasingly autocatalytic subnetworks,
shedding light on the dynamics of CRN expansion. The study offers insights into fundamental
questions about origins of chemical systems, CRN expansion over time, and the prerequisites
for generating autocatalytic species.
Bibliography
[1]
Peng, Zhen and Linderoth, Jeff and Baum, David A. The hierarchical organization of
autocatalytic reaction networks and its relevance to the origin of life. Computational
Biology, 18, 1553-7358, 2022.
[2]
Gagrani, Praful and Blanco, Victor and Smith, Eric and Baum, David. Polyhedral geometry
and combinatorics of an autocatalytic ecosystem. Journal of Mathematical Chemistry,
62, 1012–1078, 2024.
41
XIII International Workshop on Locational Analysis and Related Problems
Classification forests via mathematical programming: case study
in obesity detection
Víctor Blanco a, Alberto Japón b,*, Justo Puerto b, Peter Zhang c
aUniversity of Granada, vblanco@ugr.es
bUniversity of Seville, ajapon1@us.es puerto@us.es
cCarnegie Mellon University, pyzhang@cmu.edu
*Presenting author.
Keywords: Classification Forests, Classification Trees, Interpretable Machine Learning
In recent years, there has been an enormous development in the field of machine learning and
its applications in both academia and industry. One of the most prominent lines of research
has been interpretable machine learning, due to the general preference of practitioners to use
predictive models that are understandable from a human point of view over black-box ones
[1]. Whether models are more or less understandable for a given audience may have a certain
degree of subjectivity. However, there is agreement that the simpler the model, the better.
Mathematical optimization has made great contributions in this respect when working with
moderate sample sizes. As an example we can look at the field of classification trees, where
optimal classification trees achieve accuracy rates equal to or better than heuristic trees with
more splits [2], or at decision forests, where optimal classiication forests need fewer trees than
random forests [3].
On the other hand, in many applications, it is not only important to understand the overall
classifier solution, but also to understand certain individual cases of sample predictions when
the prediction has not been as desired. This is for example the case of a person who is denied
a mortgage loan, where it is beneficial for both the individual and the entity to understand
what the individual could change to classify as valid for the loan. These are the so-called
counterfactual explanations. In this work, we will use classification forests and mathematical
programming in a case study applied to the detection of obesity, and we will study the individual
solutions of the people in the sample to understand what can be done to prevent obesity.
Bibliography
[1]
Rudin, Cynthia, et al. Interpretable machine learning: Fundamental principles and 10
grand challenges. Statistic Surveys 16: 1-85, 2018.
43
Granada, 4-6 September 2024 IWOLOCA 2024
[2]
Bertsimas, Dimitris, and Jack Dunn. Optimal classification trees. Machine Learning, 106:
1039-1082, 2017.
[3]
Blanco, Víctor, et al. A Mathematical Programming Approach to Optimal Classification
Forests. arXiv preprint arXiv:2211.10502, 2022.
44
XIII International Workshop on Locational Analysis and Related Problems
Finding a smallest mediated set
Víctor Blanco a,b, Miguel Martínez-Antón a,b,*
a
Institute of Mathematics (IMAG), University of Granada,
vblanco@ugr.com
,
mmanton@ugr.com
bDpt. Quantitative Methods for Economics & Business, University of Granada
*Presenting author.
Keywords: Discrete Location, Mediated Sets, Sums of Squares, Circuit Polynomials
Introduction
Since Hilbert’s 17th Problem, which states if every real homogeneous polynomial (form) that
is nonnegative can be written as sum of squares (SOS) of suitable forms [2], the study of
nonnegativity of real, multivariate polynomials and sums of squares is a key problem in real
algebraic geometry and an active field of research especially due to their useful applications in
polynomial optimization developed in the last two decades.
On the one hand, a circuit polynomial is a polynomial 𝑝∈ℝ[𝑥1, . . . , 𝑥𝑑]of the form
𝑝(x)=
𝑟
Õ
𝑗=0
𝑐𝜶(𝑗)x𝜶(𝑗)+𝑐𝜷x𝜷
where
𝑟≤𝑑
, exponents
𝜶(𝑗),𝜷∈ℤ𝑑
+
, and coefficients
𝑐𝜶(𝑗)>
0and
𝑐𝜷∈ℝ∗
such that
New(𝑝)
is a simplex with even vertices
𝜶(
0
), . . . , 𝜶(𝑟)
and the exponent
𝜷
is in the strict interior of
New(𝑝).
On the other hand, a mediated set
𝐿
of an integral simplex
𝑆
with vertex,
Vert(𝑆)
, in
(
2
ℤ)𝑑
is
a subset of lattice points in ℤ𝑑∩𝑆satisfying:
•Vert(𝑆)⊂𝐿, and
•if 𝜸𝑖∈𝐿, then there exist 𝜸𝑗,𝜸𝑘∈ (2ℤ)𝑑∩𝐿with 𝜸𝑖=1
2(𝜸𝑗+𝜸𝑘).
Iliman and de Wolff [3] proved that a nonnegative circuit polynomial is SOS if and only if
there exists a mediated set
𝑀
of its Newton polytope
𝑆
that contains the exponent of one
distinguished term (
𝜷
) and, in this case, it will be a linear combination with strictly positive
coordinates of x
𝜸𝑗
2−x𝜸𝑘
22𝜸𝑖∈𝑀\Vert(𝑆).
45
Granada, 4-6 September 2024 IWOLOCA 2024
Minimum mediated set (blue dots) for
Vert(𝑆)={(
0
,
0
),(
20
,
0
),(
0
,
20
)}
(red dots), and the
interior point (2,11)(green star). Solution associated to 2𝑥20 +11 𝑦20 +7−20𝑥2𝑦11.
In [1], the authors study mediated sets in the continuous case due to their crucial role in
the minimal second-order cone representation of rational generalized power cones. They
obtain results on the combinatorial structure of these sets using mathematical optimization
tools. This approach is applied both in the continuous case to study generalized power cones
and in the discrete case to address a question arising from Hilbert’s 17th Problem. Given a
simplex
𝑆
with even vertices and a strictly interior point
𝜷
, what is the minimum mediated
set of
𝑆
that contains
𝜷
? In case this set is empty we have a certification of a nonnegative
circuit polynomial supported on
Vert(𝑆) ∪ {𝜷}
is not SOS, otherwise, we will know a minimal
representation of that polynomial as SOS, and the involved binomial.
The authors’ first approach requires enumerating every single even point in the simplex
𝑆
,
which presents a significant challenge. The next step involves finding a small set of potential
mediated points to determine the minimum mediated set. The authors propose a constructive
method to build this potential set, starting from the vertex set of the simplex and the interior
point. They also introduce suitable mathematical optimization models to find the optimal
location of the mediated points in the mediated set while enumerating as few points as
possible.
Bibliography
[1]
Blanco, V. and Martínez-Antón, M. “On minimal extended representations of generalized
power cones,” To appear in SIAM Optimization, 2024.
[2]
Hilbert, D. “Über die darstellung definiter formen als summe von formenquadraten,”
Algebra·Invariantentheorie Geometrie, 154–161, 1970.
[3]
Iliman, S. and de Wolff, T. “Amoebas, nonnegative polynomials and sums of squares
supported on circuits,” Research in the Mathematical Sciences, (3): 1–35, 2016.
46
XIII International Workshop on Locational Analysis and Related Problems
Modeling ordered interactions in Location Problems
Víctor Blanco a, Miguel A. Pozo b, Justo Puerto b, Alberto Torrejón b,*
a
Department of Quantitative Methods for Economics & Business, University of Granada,
Spain.
Institute of Mathematics of the University of Granada (IMAG), Spain,
vblanco@ugr.es,
bDepartment of Statistics and Operational Research, University of Seville, Spain.
Institute of Mathematics of the University of Seville (IMUS), Spain,
pozo@us.es,puerto@us.es,atorrejon@us.es
*Presenting author.
Keywords: Discrete location, Ordered optimization, Quadratic optimization
Introduction
Ordered optimization (see [1] & [2]) allows a straightforward and flexible generalization of
facility location problems, since many well-known problems in the literature can be modelled
using an ordered formulation. The ordered description of these problems relies on the vector
of weights, usually named
𝜆
-vector, that multiplies the ordered costs in the objective function.
By means of an appropriate
𝜆
-vector, many location problems such as median, center or
centdian problems can be modelled, but also, since negative and non-monotonous weight
vectors are allowed, one can also consider obnoxious location problems, preference modelling
or other abstract concepts such as, for example, fairness.
However, the range of problems that can be studied within an ordered framework is limited to
the linearity of the objective function. By using a quadratic approach to ordered optimization,
and considering a matrix of weights instead of a vector, the number of criteria to be modelled
is considerably multiplied, allowing, among others, to minimize interaction between costs
or to model more specific objective functions which require quadratic terms, such as the
variance. On the other hand, introducing ordered interactions as a modeling feature increases
the complexity of the problems at hand, which motivates the study of efficient resolution
methods that allow the scalability of these problems.
This work motivates this new approach, which allows a higher level of generalisation of several
combinatorial problems, particularly location problems, providing a mathematical formalization
of this approach, describing exact models as solution methods and performing empirical
comparison between these different solutions methods.
47
Granada, 4-6 September 2024 IWOLOCA 2024
Bibliography
[1] Nickel, S., Puerto, J. (2006). Location Theory: a unified approach. Springer.
[2]
Puerto, J., Rodríguez-Chía, A.M. (2019). Ordered median location problems. Location
science. Springer, pp. 249–288.
48
XIII International Workshop on Locational Analysis and Related Problems
Cover-based inequalities for the single-source capacitated facility
location problem with customer preferences
Christina Büsing a, Markus Leitner b, Sophia Wrede a,*
aRWTH Aachen University, Aachen, Germany, buesing@combi.rwth-aachen.de,
wrede@combi.rwth-aachen.de
bVrije Universiteit Amsterdam, Amsterdam, The Netherlands, m.leitner@vu.nl
*Presenting author.
Keywords: Facility location, preference constraints, valid inequalities
In the classical single-source capacitated facility location problem (CFLP), a set of facilities
needs to be chosen in order to cover the demand of customers. Customers are assigned to any
open facility such that the capacity of the facility is not exceeded and the total cost consisting
of opening and assignment costs is minimised. However, in many real-world applications
customers are not willing to travel to any open facility assigned to them but want to select an
open facility according to their preferences [2, 3]. Such deviations can turn feasible solutions
for the CFLP infeasible. The single-source capacitated facility location problem with customer
preferences (CFLP-CP) takes this behavior into account by assigning each customer to their
most preferred open facility.
Both the CFLP and CFLP-CP are strongly NP-hard. Preference constraints, however, imply
certain combinatorial structures which do not occur in the classical CFLP. These implied
structures render some previously NP-hard special cases of the CFLP easy to solve[1].
One remarkable structure implied by preference constraints intertwines the assignments of
customers with related preferences to the same facility.
In this talk, we focus on cover-based inequalities for the CFLP-CP. We first strengthen the
well-known cover inequalities so that they incorporate the specific structure of the CFLP-CP
mentioned above. Afterwards, we strengthen these inequalities by utilising further properties
of solutions for the CFLP-CP arising from the combination of capacities and preference
constraints. Lastly, we evaluate the impact of these considered inequalities for two preference
types in a computational study.
49
Granada, 4-6 September 2024 IWOLOCA 2024
Bibliography
[1]
Büsing, C., Gersing, T., and Wrede, S. Analysing the complexity of facility location
problems with capacities, revenues, and closest assignments. Proceedings of the 10th
International Network Optimization Conference (INOC), Aachen, Germany, June 7-–10,
pages 81—86, 2022.
[2]
Kang, C.-N., Kung, L.-C., Chiang, P.-H., and Yu, J.-Y. A service facility location problem
considering customer preference and facility capacity. Computers & Industrial Engineering,
Volume 177, 2023.
[3]
Polino, S., Camacho-Vallejo, J.-F., and Villegas, J.G. A facility location problem for
extracurricular workshop planning: bi-level model and metaheuristics. Intl. Trans. in Op.
Res., 2024.
50
XIII International Workshop on Locational Analysis and Related Problems
A bi-level metaheuristic for the design of hierarchical transit net-
works in a multimodal context
David Canca a,*, Xingrong Wang b
aUniversidad de Sevilla, dco@us.es
bSchool of Systems Science, Beijing, 20114009@bjtu.edu.cn
*Presenting author.
Keywords: Network Design, Hierarchical transportation network, ALNS
Introduction
In many large cities around the world, rapid rail rapid transit systems are used as the main
transportation mode to meet the mobility needs of citizens. Despite the large volume of
passengers transported, rapid transit lines rarely can meet the total demand for passengers.
The construction of a well-coordinated metro-bus bimodal system may benefit not only
passengers performing bimodal trips in the city area, those who see their travel possibilities
increased, but also passengers using each individual transit mode, since sharing the total
demand, the congestion of both modes can be easily managed. If both the metro and bus
regard themselves as a self-standing system, an uncoordinated metro-bus bimodal configuration
with excessive intermodal competition or defective intermodal connection may occur, just like
the dilemma faced by the metro and bus systems of some megacities in China. To ensure a
safe operation, it has been necessary to implement different passenger demand management
measures, such as passenger inflow control systems ([1, 2, 3]. However, the bus demand
has decreased year after year. Many bus operators have to rely on unsustainable government
financial subsidies to maintain daily operations [4]. Since modifying the bus network is relatively
easy and cheaper than acting over the railway network, renewing the bus network design
to adapt it to the existing metro system is the most efficient way to improve the bi-modal
network functionality. Although several previous studies investigated the effect of the metro
on a bus network design taking into account the modal choice for passengers (e.g., [4, 5]), no
specific attention has been paid to the potential effect of considering different types of bus
lines, forming a hierarchical system, working together with an existing metro network. To
compensate for the deficiencies existing in both actual operation and theoretical research,
this research deals with the problem of simultaneously determining the configuration of a
hierarchical bus network and the frequency of the lines to ensure a coordinated operation
with an existing metro, explicitly considering the multimodal route choice of passengers. The
51
Granada, 4-6 September 2024 IWOLOCA 2024
hierarchical bus network structure consists of different line types: regular, trunk, and feeder
lines. Each type corresponds to a specific stopping pattern -characterized by a particular
inter-stop distance, commercial speed, and terminal stations-. The problem is formulated as a
non-convex non-linear optimization model. To efficiently solve the model, a bi-level heuristic
that breaks down the integrated problem into two interlaced subproblems and then solves
them iteratively is presented. In the upper level, an adaptive large neighbourhood search
metaheuristic is responsible for proposing new network designs, whereas, in the lower level, a
heuristic procedure attempts to determine the best frequencies and perform the passenger
assignment. The obtained results demonstrate the capabilities of the proposed approach. As
main conclusion, by achieving a trade-off between lines with different running speeds resulting
from different average inter-stop distances, hierarchical bus network designs can incentivize
commuters into trip plans that contribute to improving the overall integrated metro-bus
bimodal network performance.
Bibliography
[1]
Wang, X., Wu, J., Yang, X., Guo, X., Yin, H., Sun, H. Multistation coordinated and
dynamic passenger inflow control for a metro line. IET Intelligent Transport Systems. 14,
1068–1078, 2020.
[2]
Zhang, P., Sun, H., Qu, Y., Yin, H., Jin, J.G., Wu, J. Model and algorithm of coordinated
flow controlling with station-based constraints in a metro system. Transportation Research
Part E: Logistics and Transportation Reviews. 148, 102274, 2021.
[3]
Hu, Y., Li, S., Wang, Y., Zhang, H., Wei, Y., Yang, L. Robust metro train scheduling
integrated with skip-stop pattern and passenger flow control strategy under uncertain
passenger demands. Computers & Operations Research. 151, 106116, 2023.
[4]
Liang, J., Wu, J., Gao, Z., Sun, H., Yang, X., Lo, H.K. Bus transit network design with
uncertainties on the basis of a metro network: A two-step model framework. Transportation
Research Part B: Methodological. 126, 115–138, 2019.
[5]
Wan, Q.K., Lo, H.K. Congested multimodal transit network design. Public Transport. 1,
233–251, 2009.
52
XIII International Workshop on Locational Analysis and Related Problems
Novel Valid Inequalities for Chance-Constrained Problems with
Finite Support
Diego Cattaruzza
a,b
, Martine Labbé
c,b,*
, Matteo Petris
b
, Marius Roland
d
, Martin
Schmidt e
aCentrale Lille, diego.cattaruzza@centralelille.fr
bINOCS Team, INRIA, Lille, matteo.petris@inria.fr
cUniversité Libre de Bruxelles, martine.labbe@ulb.be
dPolytechnique Montreal, mmmroland@gmail.com
eTrier University, martin.schmidt@uni-trier.de
*Presenting author.
Keywords: Chance constraints, MILP, Valid inequalities
Introduction
We consider the (mixed-integer) linear chance-constrained problem
min
𝑥𝑐⊤𝑥(1a)
s.t. 𝑥∈𝑋, (1b)
ℙ(𝐴𝜔𝑥−𝑏𝜔≥0) ≥ 1−𝜏(1c)
where
𝑋⊂ℝ𝑛
is a non-empty and compact set defined by deterministic constraints, which
possibly include integrality restrictions on some or all of the variables. The cost vector is given
by
𝑐∈ℝ𝑛
. Let
(Ω,
2
Ω,ℙ)
be a discrete and finite probability space such that
Ω = {𝜔𝑠
:
𝑠∈ 𝒮}
and
𝑝𝑠= ℙ(𝜔=𝜔𝑠)
for
𝑠∈ 𝒮
and
Í𝑠∈𝒮 𝑝𝑠=
1holds. Moreover, let
𝐴𝜔∈ℝ𝑚×𝑛
be a matrix
of random constraint coefficients and let
𝑏𝜔∈ℝ𝑚
be a vector of random right-hand sides.
Finally, 𝜏∈ (0,1)is the risk tolerance.
It is well known that this problem can be reformulated as a Mixed-Integer Linear Problem
(MILP) by introducing one binary variable
𝑦𝑠
for each scenario such that
𝑦𝑠
takes value 1if
the constraint set of that scenario is satisfied by the solution.
We derive two novel families of valid inequalities. One is based on strong duality of the linear
programming formulation of the respective quantile, whereas the other exploits a covering
argument. We prove that the covering-based inequalities are stronger than the first family as
well as that they dominate the covering inequalities proposed in [1] for the case of a single
53
Granada, 4-6 September 2024 IWOLOCA 2024
constraint per scenario. Moreover, we state and prove results on the closure of these valid
inequalities and compare the closure of our second family of inequalities with the closure of
the so-called mixing-set inequalities introduced in [2].
Finally, we illustrate the impact of our novel valid inequalities in a numerical study based on
instances proposed in [3].
Bibliography
[1]
Qiu, F. and Ahmed, S. and Dey, S. S. and Wolsey, L. A. Covering linear programming
with violations. INFORMS Journal on Computing, 26(3), 531-546, 2014.
[2]
Luedtke, J. A branch-and-cut decomposition algorithm for solving chance-constrained
mathematical programs with finite support. Mathematical Programming, 146(1-2),
219-244, 2014.
[3]
Song, Y. and Luedtke, J. R. and Küçükyavuz, S. Chance-constrained binary packing
problems. INFORMS Journal on Computing, 26(4), 735-747, 2014.
[4]
Ahmed, S. and Luedtke, J. and Song, Y. and Xie, W. Nonanticipative duality, relaxations,
and formulations for chance-constrained stochastic programs. Mathematical Programming,
162, 51-81, 2017.
54
XIII International Workshop on Locational Analysis and Related Problems
Stable Solutions for the Capacitated Simple Plant Location Prob-
lem with Order
Concepción Domínguez a,*
aUniversity of Murcia, concepcion.dominguez@um.es
*Presenting author.
Keywords: Location with preferences, Simple Plant Location Problem, Facility Location,
Combinatorial Optimization, Matchings under preferences
Introduction
The Simple Plant Location Problem (SPLP) is a well-known problem in the Location Field
where the aim is to open a set of facilities and allocate each customer to a facility minimizing
the total cost of location plus allocation. In the SPLP, it is assumed that the customers have
no decision in the allocation, so they are allocated to the open plant that minimizes the cost.
The SPLP with Order (SPLPO) arises when the preferences of the customers regarding their
allocation is taken into account in the allocation process. Thus, in the SPLPO customers
rank the facilities according to their preferences and they are allocated to their most-preferred
facility among the open ones. The SPLPO was introduced in 1987 [4] and has since been
thoroughly studied in the literature (see. e.g. [1, 3]).
In the Capacitated version of the problem (CSPLPO) it is assumed that there is a limited
number of customers that can be allocated to each plant. When capacities in the plants and
customers’ preferences are involved at a time, a decision rule has to be taken to discern which
customers are allocated to plants with high demand. As might be expected, any solution with
full plants can have envy customers, i.e. customers that would have liked to be allocated to a
different plant which is full. Different decision rules on how to deal with the allocation result
in different versions of the problem. Very recently, a bilevel problem has been proposed in [2]
where the ranking of each customer is given as a list of numerical preferences (the smaller the
value, the greater the preference towards the plant) and the aim of the follower problem is to
maximize the preferences of the customers globally.
In our work, we propose a new approach where only a strict total order of the plants (with
no numerical values) is taken into consideration for each customer. In this setting, new
decision criteria for the allocation are defined based on the concept of stability in an allocation.
New formulations are proposed for each decision criteria, along with valid inequalities and
theoretical results designed to obtain compact and tighten them. Preliminary computational
55
Granada, 4-6 September 2024 IWOLOCA 2024
results showing the efficiency of the approach and the resultant stable solutions are provided.
Bibliography
[1]
Cabezas, X. and García, S. A semi-Lagrangian relaxation heuristic algorithm for the simple
plant location problem with order. Journal of the Operational Research Society, 74(11),
2391-2402, 2023.
[2]
Calvete, H. I., Galé, C., Iranzo, J. A., Camacho-Vallejo, J. F. and Casas-Ramírez, M.
S. A matheuristic for solving the bilevel approach of the facility location problem with
cardinality constraints and preferences. Computers & Operations Research, 124, 105066,
2020.
[3]
Cánovas, L., García, S., Labbé, M. and Marín, A. A strengthened formulation for the
simple plant location problem with order. Operations Research Letters, 35(2), 141-150,
2007.
[4]
Hanjoul, P. and Peeters, D. A facility location problem with clients’ preference orderings.
Regional Science and Urban Economics, 17(3), 451-473, 1987.
56
XIII International Workshop on Locational Analysis and Related Problems
Cross-docking platforms design and distributionally robust opti-
mization
Laureano F. Escudero a,* , María Araceli Garín b, Aitziber Unzueta c
a
Statistics and Operations Research Area, Universidad Rey Juan Carlos (URJC), Móstoles
(Madrid), Spain, laureano.escudero@urjc.es
b
Quantitative Methods Dep., Universidad del País Vasco (UPV/EHU), Bilbao (Bizkaia),
Spain, mariaaraceli.garin@ehu.eus
c
Applied Mathematics Dep., Universidad del País Vasco (UPV/EHU), Bilbao (Bizkaia), Spain,
aitziber.unzueta@ehu.eus
*Presenting author.
Keywords: Cross-dock door design, two-stage stochastic quadratic combinatorial optimiza-
tion, Distributionally robust optimization, Ambiguity sets
The Cross-dock Door Design Problem (CDDP) consists of deciding on the number and
capacity of inbound and outbound doors, minimizing the construction and exploitation cost of
the infrastructure. The uncertainty, realized in scenarios, lies in the occurrence of these nodes,
the delivering material volume and cost, and the capacity’s disruption of the doors. Introducing
a scheme for generating the ambiguity sets for the second stage uncertain parameters in
CDDP.
Introduction
Distributionally robust optimization (DRO) is motivated as a counterpart of the usually
unknown underlying probability distribution (PD)followed by the uncertainty in dynamic
problems. This wok presents a DRO scheme for generating an ambiguity set to represent
the uncertainty in a two-stage scenario tree for the highly combinatorial CDDP to decide the
number and nominal capacity of the strip and stack doors.
It is assumed the availability of a Nominal Distribution (ND) for the realization of the uncertain
parameters in the second stage of the tree. That ambiguity set is generated by considering
the projections of appropriate perturbations of the cumulative distribution functions of the
ND realizations in a set of modeler-driven PDs (say Normal, Weybull, Gamma and Lognormal
distributions). Those perturbations are selected based on the minimization of the Wasserstein
distance between the parameters’ realizations in each member in the ambiguity set and the
ND realization, so that a given radius is satisfied.
The underline assumption is that the realizations of the uncertain parameters in modeler-driven
57
Granada, 4-6 September 2024 IWOLOCA 2024
groups are identically distributed random variables, being independent from the parameters
that belong to other groups.
Bibliography
[1]
Gao, R. & Kleywegt, A.J. (2022). Distributionally robust stochastic optimization with
Wasserstein distance. Mathematics of Operations Research, 48:603-655. Previously,
ArXiv:1604.02199c2, 2016.
[2]
Kantorovich, L. (1939). The mathematical method of production planning and organization.
Management Science, 6:366-422.
[3]
Morales, D. (2022) On projections of cdf perturbations of multiple variables in probabilistic
distributions. Private conversation.
[4] Pardo, L (2005) Statistical inference based on divergence measures. CRC Press.
[5]
Scarf, H.E. (1957). A min-max solution of an inventory problem. Technical report P-910,
Rand Corporation, Santa Monica, CA, USA.
58
XIII International Workshop on Locational Analysis and Related Problems
A further research on Stochastic Single-Allocation Hub Location
Problem
Inmaculada Espejo
a,*
, Alfredo Marín
b
, Juan M. Muñoz-Ocaña
a
, Raúl Páez
a
, Antonio
M. Rodríguez-Chía a
a
Departamento de Estadística e Investigación Operativa, Universidad de Cádiz, Spain,
inmaculada.espejo@uca.es,juanmanuel.munoz@uca.es,raul.paez@uca.es,
antonio.rodriguezchia@uca.es
b
Departamento de Estadística e Investigación Operativa, Universidad de Murcia, Spain,
amarin@um.es
*Presenting author.
Keywords: Stochastic, Hub, Branch-and-cut
Introduction
The Hub Location Problem (HLP) has been addressed by a wide community of operations
researchers due to its practical relevance. Many reviews about location problems show the
extensive activity in this field and the applications of these problems, see [1], [2] and [3],
among others.
Most studies on the HLPs are focused on models where the input parameters are assumed
to be fixed and known. However, in most real-life problems, parameters such as demands,
transportation costs, capacity of hubs, et cetera are subject to variation due to a variety
of factors such as population shifts, economic, environmental and political circumstances,
quality of the provided services, et cetera. Hence, some information required for planning is
not available a priori, and the associated uncertainties are only resolved once the system is
constructed and the hubs are installed.
This work deals with a Single-Allocation Hub Location Problem (SAHLP) where the amount
of product sent between origins and destinations, the transportation costs and the capacities
of the hubs, are uncertain and modelled using random variables with realization only after
the hubs are selected. Different scenarios are considered (with their probabilities) and two
decisions have to be made: i) the location of the hubs (this location does not change from one
scenario to another), and ii) the allocation of each origin/destination to these hubs in order
to minimize the expected overall costs of the system. This is a realistic practice because the
hubs are located before knowing the real scenario and the allocations are determined when the
actual parameters are realized. This is motivated by the necessity of the network operators to
59
Granada, 4-6 September 2024 IWOLOCA 2024
quickly react to the changes in overall system performance by adjusting the assignments of
the origins/destinations to the hubs when the uncertainty is realized.
A compact integer programming formulation is proposed for the stochastic SAHLP with three
variants: uncapacitated, capacitated and
𝑝
-hub problem. Valid inequalities are developed to
reinforce the formulation and added in a branch-and-cut procedure.
Bibliography
[1]
Alumur, S., Campbell, J. F., Contreras, I., Kara, B., Marianov, V., and O’Kelly, M. E.
Perspectives on modeling hub location problems. European Journal of Operational
Research, 291(1), 1-17, 2021.
[2]
Campbell, J. F. and O’Kelly, M. E. (2012). Twenty-five years of hub location research.
Transportation Science, 46(2), 153-169, 2012.
[3]
Contreras, I. and O’Kelly, M. E. Hub location problems. In Laporte, G., Nickel, S., and
Saldanha-da-Gama, F., editors, Location Science, 327-363. Springer, New York, 2019.
60
XIII International Workshop on Locational Analysis and Related Problems
A Reformulation for the Medianoid Problem with Multipurpose
Trips
Juan Pablo Fernández-Gutiérrez
a
, Juan G. Villegas
b
, José-Fernando Camacho-Vallejo
c,*
aFaculty of Basic Science, Universidad de Medellín, Colombia,
jpfernandez@udemedellin.edu.co
b
ALIADO-Analytics and Research for Decision Making, Department of Industrial Engineering,
Universidad de Antioquia, Colombia, juan.villegas@udea.edu.co
cTecnologico de Monterrey, Escuela de Ingeniería y Ciencias, Mexico,
fernando.camacho@tec.mx
*Presenting author.
Keywords: Competitive Facility Location, Bilevel Optimization, Medianoid Problem, Travel-
ling Purchaser Problem
Introduction
Competitive Facility Location Problems (CFLPs) involve facilities from different companies
offering similar commodities or services. Typically, it is assumed that some facilities are
already operational in the market, and new ones will be established to increase or maintain
companies’ market share [1]. Due to its similarity to Stackelberg games, CFLPs often employ
leader-follower terminology. Within this hierarchical framework, the leader initiates the process
by locating their new facilities. Subsequently, taking the leader’s decisions into account, the
follower determines the locations of their own facilities.
These decisions influence the allocation of customers, which often follows a binary assignment
rule. Typically, either the closest distance or the cheapest assignment criteria are used. In our
study, we assume that the leader already operates some facilities in the market. Therefore, we
are examining the location problem from the follower’s perspective, with the aim of maximizing
its market share.
Motivation
The motivation for this study stems from the observation that much of the existing research
on CFLPs assumes single-purpose trips. In such scenarios, customers typically make back-and-
61
Granada, 4-6 September 2024 IWOLOCA 2024
forth trips to facilities to pick up their demanded commodities. However, when multipurpose
trips are considered, customers strategically plan their trips to purchase commodities from
different locations.
Incorporating routing decisions into the facility location decision-making process is crucial to
prevent suboptimal solutions in practice [2].
The Medianoid Problem with Multipurpose Trips
In this research, we address the Medianoid Problem with Multipurpose Trips (MPMT) as
proposed in [3]. In this problem, a company enters the market (referred to as the follower in
the CFLP context) with the objective of locating
𝑟
facilities to maximize its market share
(capturing customer demand). Customers have associated demand for various commodities,
which can be fulfilled from different facilities. Subsequently, customers determine their routing
for collecting their demanded commodities from both existing and newly located facilities.
In this problem, the entering firm acts as the leader, while the customers serve as the followers.
To solve the problem, we present a reformulation of the bilevel problem, establishing an
equivalence with the maximal covering location problem. This reformulation enables exact
and efficient problem-solving. Computational experimentation validates this approach.
Bibliography
[1]
Mishra, Mamta and Singh, Surya Prakash and Gupta, MP. Location of competitive facilities:
a comprehensive review and future research agenda. Benchmarking: An International
Journal, 2022.
[2]
Salhi, Said and Rand, Graham K. The effect of ignoring routes when locating depots.
European journal of operational research, 39(2), 150-156, 1989.
[3]
Khapugin, Sergey and Melnikov, Andrey Local search approach for the medianoid problem
with multi-purpose shopping trips. Proceedings of the International Conference on
Mathematical Optimization Theory and Operations Research, 328-341, 2019.
62
XIII International Workshop on Locational Analysis and Related Problems
Solving a multi-source capacitated facility location problem with
a fairness objective
Carlo Filippi a, Gianfranco Guastaroba a, Juan-José Salazar-González b,*
aDEM, University of Brescia, Brescia, Italy,
carlo.filippi,gianfranco.guastaroba}@unibs.it
bIMAULL, University of La Laguna, La Laguna 38200, Tenerife, Spain jjsalaza@ull.es
*Presenting author.
Keywords: Location, Fairness, Cutting plane methods
Introduction
In several application domains, a decision-maker has to locate a set of facilities, which provide
a service (or good), and that are reached by “customers”, usually individuals, at their own costs.
In these situations, improving system efficiency requires minimizing the average cost paid by
customers to reach a facility, while improving fairness requires minimizing the variability in
the cost distribution [2]. Further, efficiency and fairness are commonly recognized as two
competing objectives, where the former is usually the primary objective in the private sector,
whereas the latter is normally the primary objective in the public sector [1].
In this paper, we measure accessibility fairness by using the conditional
𝛽
-mean [5], an equity
measure strictly related to the Conditional Value-at-Risk, a popular risk measure [3].
As long as single-source location problems are involved (i.e., problems where the demand
from a customer cannot be split among different facilities), the conditional
𝛽
-mean can be
incorporated into a MILP model preserving linearity [2]. On the contrary, in multi-source
location problems where the demand of a given customer can be allocated to more than one
facility, incorporating the conditional
𝛽
-mean into a MILP may lead to mixed integer non-linear
programs. An example can be found in [4].
Our contributions can be summarized as follows: (1) we provide a non-linear formulation
for the fair multi-source capacitated facility location problem; (2) we propose two linear
reformulations of the latter model by using ideas from bilevel programming; (3) we develop
solution methods that work along the general lines of a cutting-plane algorithm; (4) we show
the effectiveness of the proposed algorithms.
63
Granada, 4-6 September 2024 IWOLOCA 2024
Models
The multi-source capacitated facility location problem is formulated as follows:
(MSCFL)min Õ
𝑗∈𝐽
𝑓𝑗𝑦𝑗+Õ
𝑖∈𝐼Õ
𝑗∈𝐽
𝑐𝑖 𝑗 𝑑𝑖𝑥𝑖𝑗 (2)
subject to Õ
𝑗∈𝐽
𝑥𝑖𝑗 =1(𝑖∈𝐼)(3)
Õ
𝑖∈𝐼
𝑑𝑖𝑥𝑖𝑗 ≤𝑠𝑗𝑦𝑗(𝑗∈𝐽)(4)
𝑦𝑗∈ {0,1} (𝑗∈𝐽)(5)
𝑥𝑖𝑗 ≥0(𝑖∈𝐼;𝑗∈𝐽).(6)
where
𝐼
is the set of customers,
𝐽
is the set of potential facilities,
𝑑𝑖
is the demand of customer
𝑖∈𝐼
,
𝑓𝑗
and
𝑠𝑗
are, respectively, the fixed cost and the capacity of facility
𝑗∈𝐽
, and finally
𝑐𝑖𝑗 is the distance between 𝑖and 𝑗. Let 𝒳𝒴 denote the feasible set of MSCFL.
Definition 1. Given a parameter
𝛽∈ (
0
,
1
]
and a vector
(
x
,
y
) ∈ 𝒳𝒴
, the conditional
𝛽
-mean
𝑀𝛽(
x
)
is the average distance travelled by the 100
×𝛽
percent of total demand that travels
the longest distance to reach the assigned facility.
The conditional 𝛽-mean for a MSCFL problem can be expressed as:
𝑀𝛽(x)=min
𝐷𝛽𝑢+Õ
𝑖∈𝐼Õ
𝑗∈𝐽
𝑑𝑖𝑣𝑖𝑗 𝑥𝑖 𝑗 𝑢+𝑣𝑖𝑗 ≥𝑐𝑖 𝑗
𝐷𝛽
;𝑣𝑖𝑗 ≥0(𝑖∈𝐼 , 𝑗 ∈𝐽)
(7)
where
𝐷𝛽=𝛽·Í𝑖∈𝐼𝑑𝑖
. Problem (7) can be embedded into
MSCFL
, obtaining a non-linear
formulation for the fair
MSCFL
problem. We linearise such a formulation by transforming it
into an equivalent bilevel program, and then employing standard techniques to reduce the
bilevel program into a single-level optimization program. In this way, we produce two different
formulations and discuss computational results.
Bibliography
[1]
Eiselt, H.A., Laporte, G.: Objectives in location problems. In: Drezner, Z. (ed.) Facility
Location: A Survey of Applications and Methods, pp. 151–180. Springer New York (1995)
[2]
Filippi, C., Guastaroba, G., Speranza, M.G.: On single-source capacitated facility location
with cost and fairness objectives. European Journal of Operational Research 289(3),
959–974 (2021)
[3]
Filippi, C., Ogryczak, W., Speranza, M.G.: Bridging
𝑘
-sum and CVaR optimization in
MILP. Computers & Operations Research 105, 156–166 (2019)
64
XIII International Workshop on Locational Analysis and Related Problems
[4]
Givler Chapman, A., Mitchell, J.E.: A fair division approach to humanitarian logistics
inspired by conditional value-at-risk. Annals of Operations Research 262, 133–151 (2018)
[5]
Ogryczak, W., Zawadzki, M.: Conditional median: A parametric solution concept for
location problems. Annals of Operations Research 110, 167–181 (2002)
65
XIII International Workshop on Locational Analysis and Related Problems
Cross-docking platforms design and mixed binary quadratic model
for distributionally robust optimization
María Araceli Garín a,*, Laureano F. Escudero b, Aitziber Unzueta c
a
Quantitative Methods Dep., Universidad del País Vasco (UPV/EHU), Bilbao (Bizkaia),
Spain, mariaaraceli.garin@ehu.eus
b
Statistics and Operations Research Area, Universidad Rey Juan Carlos (URJC), Móstoles
(Madrid), Spain, laureano.escudero@urjc.es
c
Applied Mathematics Dep., Universidad del País Vasco (UPV/EHU), Bilbao (Bizkaia), Spain,
aitziber.unzueta@ehu.eus
*Presenting author.
Keywords: Cross-dock door design, stochastic two-stage scenario setting, mixed binary
quadratic, distributionally robust optimization
The Cross-dock Door Design Problem (CDDP) consists of deciding on the number and
capacity of inbound and outbound doors, to minimize the construction and exploitation cost
of the infrastructure. The uncertainty, realized in a two-stage scenario setting, lies in the
occurrence of these nodes, the delivering material volume and cost, and the possible capacity’s
disruption of the doors. A mixed 0-1 quadratic model for distributionally robust optimization
is introduced.
Introduction
Distributionally robust optimization (DRO) is motivated as a counterpart of the usually
unknown underlying probability distribution (PD) followed by the uncertainty in dynamic
problems. An approach is presented for the highly combinatorial Cross-dock Door Design
Problem (CDDP) solving to decide the number and nominal capacity of the strip and stack
doors. It is assumed that the uncertainty is represented in the second stage of a two-stage
setting, where a set of scenarios for the uncertain parameters is considered for each member
of an ambiguity set that has been previously generated.
Initially, a mixed binary quadratic DRO risk-neutral (RN) model is presented. The objective
function to minimize is included by the the cross-dock infrastructure building cost (i.e., first
stage door selection and installation cost) plus the highest second stage assignment expected
cost in the scenarios, among the ambiguity set members. The constraint system is composed
of the first stage constraints and the second stage constraint set for each member of the
ambiguity set.
67
Granada, 4-6 September 2024 IWOLOCA 2024
In a second step, a risk-averse (RA) formulation is developed by adding the system of
constraints and variables related to the risk-averse measure called second-order stochastic
dominance to prevent, up to some extent, the negative implication of black swan scenarios in
the value of the objective function to minimize.
Given the difficulty on solving this highly combinatorial problem, a mathematically equivalent
MILP formulation is considered. Since, even this formulation is still difficult to solve, a scenario
Cluster Lagrangean Decomposition (CLD) is introduced for obtaining lower bounds to the
MILP model. A lazy scheme for obtaining feasible solutions to the original model based on the
CLD one is considered, by exploiting the special structure of the problem. A computational
study validates the proposal.
Bibliography
[1]
Bayraksan, G., Maggioni, F., Faccini, D. & Yang, M. (2024). Bounds for multistage
mixed-integer distributionally robust optimization. SIAM Journal on Optimization, 34:682-
717.
[2]
Gao, R. & Kleywegt, A.J. (2022). Distributionally robust stochastic optimization with
Wasserstein distance. Mathematics of Operations Research, 48:603-655. Previously,
ArXiv:1604.02199c2, 2016.
[3]
Pflug, G.C. (2023). Multistage stochastic decision problems: Approximation by recursive
structures and ambiguity modeling. European Journal of Operational Research, 306:1027-
1037.
[4]
Scarf, H.E. (1957). A min-max solution of an inventory problem. Technical report P-910,
Rand Corporation, Santa Monica, CA, USA.
68
XIII International Workshop on Locational Analysis and Related Problems
Formulations and resolution procedures for upgrading hub net-
works
Mercedes Landete Ruiz
a
, Juan M. Muñoz-Ocaña
b,*
, Antonio M. Rodríguez-Chía
b
,
Francisco Saldanha-da-Gama c
aUniversidad Miguel Hernández, landete@umh.es
bUniversidad de Cádiz, juanmanuel.munoz@uca.es,antonio.rodriguezchia@uca.es
c
Sheffield University Management School,
francisco.saldanha-da-gama@sheffield.ac.uk
*Presenting author.
Keywords: Hub location, Edge upgrading
Introduction
Hub location problems have become an important stream of research within Location Science
due to their relevance in many applications emerging in logistics, telecommunications, and
transportation [1]. This talk presents different formulations for the single-allocation hub
location problem with edge upgrades. In the broad context of network design and optimization,
upgrading aims to reduce the transportation costs between sites that are connected by the
upgraded edges. Two types of connections are considered to be upgraded: inter-hubs edges
and edges that connect origin/destination sites to hubs.
Upgrading may be motivated by different contexts. In telecommunications, upgrading an edge
may involve using a higher speed cable between a pair of servers. Similarly, in logistics and
transportation, upgrading an edge may correspond to using a faster transportation mode,
such as using a plane instead of a truck between a pair of cities, thus achieving a reduced
travel time. Another possibility is to redesign a trajectory (e.g. using a highway instead of a
secondary road).
Formulations for solving the problem
We consider a maximum number of connections that can be upgraded between hubs and a
maximum number of connections between spokes (non-hubs) and hubs. This may correspond,
for instance, to a limited number of alternative vehicles available for accomplishing the
improvement.
We start by considering complete hub networks. Afterward, we extend the work to problems
in which hub network design decisions have also to be made. For the latter, no specific
topology is imposed for the hub-level network. Given that the objective function accounts
69
Granada, 4-6 September 2024 IWOLOCA 2024
only for the transportation costs, a constraint is set on the number of hubs to install [2].
These formulations present high computing burden, and therefore, various valid inequalities
are included as cuts in branch-and-cut procedures. The methodological developments are
computationally tested for two purposes: firstly, to emphasize the relevance of embedding
upgrading decisions in single-allocation hub location problems by analyzing the savings achieved
and the structural changes in the hub networks after upgrading edges; secondly, to evaluate
the effectiveness of the formulations and solution methods for different instances.
Bibliography
[1]
Alumur, S., Campbell, J.F., Contreras, I., Kara, B., Marianov, V. and O’Kelly, M.E.
Perspectives on modeling hub location problems. European Journal of Operational
Research, 291(1), 1–17, 2021
[2]
Contreras, I. and O’Kelly, M. Hub Location Problems. Location Science. Springer Nature
Switzerland, 327-363, 2019
70
XIII International Workshop on Locational Analysis and Related Problems
Synchronization routing for agricultural vehicles and implements
Aitor López-Sánchez a,b,*, Marin Lujak a, Frédéric Semet b, Holger Billhardt a
aCETINIA, University Rey Juan Carlos, Madrid, Spain, aitor.lopez@urjc.es,
marin.lujak@urjc.es,holger.billhardt@urjc.es
bUMR 9189 - CRIStAL, Centrale Lille, Univ. Lille, CNRS, Inria, France,
frederic.semet@centralelille.fr
*Presenting author.
Keywords: Routing, Synchronization, Column generation, Agriculture
Introduction
Agriculture technology is experiencing a revolution to enhance efficiency, sustainability, and
productivity. Autonomous agriculture mobile robots (agribot) are increasingly used. Current
approaches for Agricultural Vehicle Routing Problems (AVRPs) focus on homogeneous fleets,
dealing with a single type of task and crop [5]. In contrast, real-world farming requires
coordination between different tractors/agribots and implements to manage different crops
and tasks, such as plowing, fertilizing, and harvesting. The Agricultural Fleet Vehicle Routing
Problem with Implements (AFVRPI) coordinates the routes for mixed fleets, optimizing the
cost for covering task execution and synchronization in modern agriculture.
Movement synchronization between two vehicle classes (tractors and implements) is necessary,
where the tractors have autonomous movement capacity and implements depend on the
tractors to move [3]. Various approaches in the literature address similar challenges. The
Vehicle Routing Problem with Trailers and Transshipments [2] allows for the detachment
and reattachment of trailers between trucks. The Active Passive Vehicle Routing Problem
introduces a scenario with active and passive vehicles, where the active vehicles displace the
passive ones [4]. Lastly, one of the first movement synchronization constraints in precision
agriculture is the Synchronized Sprayer Tanker Routing Problem with Variable Service Time,
which involves coordination between tender tankers and sprayer fleets [1].
Problem definition and solution approach
The Agricultural Fleet Vehicle Routing Problem with Implements (AFVRPI) is composed of a
fleet comprising both tractors and implements (ℱ =𝒱 ∪ ℳ), and their compatibilities with
each other. The transportation network is modeled by a directed graph
(𝒩 ,𝒜)
. Nodes consist
of four distinct sets:
𝒩𝑡𝑎𝑠𝑘 𝑠
agricultural tasks,
𝒩𝑑𝑒𝑝𝑜𝑡𝑠
tractor and implement depots,
𝒩𝑑𝑒𝑡𝑎𝑐ℎ
71
Granada, 4-6 September 2024 IWOLOCA 2024
detaching nodes and
𝒩𝑎𝑡𝑡𝑎𝑐ℎ
attaching nodes. Each task has a given demand, service time, and
time window. Arcs denote spatial and temporal connectivity and let us denote transfer arcs
as
(𝑑, 𝑎) ∈ 𝒜𝑡 𝑟𝑎𝑛𝑠 𝑓 𝑒𝑟
, including arcs where implements can be detached
𝑑∈ 𝒩𝑑𝑒𝑡𝑎𝑐ℎ
/attached
𝑎∈ 𝒩𝑎𝑡𝑡𝑎𝑐ℎ
to tractors. The goal of the AFVRPI is to find a set of compatible routes for
tractors and implements that visit all the tasks minimizing the overall cost and respecting the
movement constraints.
The proposed solution approach is composed of an extended master problem formulation
with independent subproblems associated with each tractor and implement, and solved with
column generation, which enables distributed and asynchronous problem-solving. Tractor
subproblems are the Elementary Shortest Path Problem with Resource Constraints (ESPPRC)
incorporating linear costs [4] with two resources: the distance, restricted by vehicle autonomy,
and the time, constrained by task time windows. Implement subproblems are also the ESPPRC
considering only demand constraints, limited by its capacities. Preliminary computational
results demonstrate the efficiency of this approach.
Bibliography
[1]
Alkaabneh, F. Matheuristic for synchronized vehicle routing problem with multiple
constraints and variable service time: Managing a fleet of sprayers and a tender tanker.
Computers & Operations Research, 162, 106454, 2024.
[2]
Drexl, M. Applications of the vehicle routing problem with trailers and transshipments.
European Journal of Operational Research, 227(2), 275-283, 2013.
[3]
Soares, R. and Marques, A. and Amorim, P. and Parragh, S. N. Synchronisation in vehicle
routing: classification schema, modelling framework and literature review. European
Journal of Operational Research, 313(3), 817-840, 2023.
[4]
Tilk, C. and Bianchessi, N. and Drexl, M. and Irnich, S. and Meisel, F. Branch-and-price-
and-cut for the active-passive vehicle-routing problem. Transportation Science, 52(2),
300-319, 2018.
[5]
Utamima, A. and Djunaidy, A. Agricultural Routing Planning: A Narrative Review of
Literature. Procedia Computer Science, 197, 693-700, 2022.
72
XIII International Workshop on Locational Analysis and Related Problems
The Effect of Budget Limiting on the Linear Ordering Problem
Inigo Martin Melero a,b,* , Mercedes Landete Ruiz b, Javier Alcaraz Soria b
aEuropean Organisation for Nuclear Research (CERN), inigo.martin.melero@cern.ch
b
Universidad Miguel Hernandez de Elche,
inigo.martin@goumh.umh.es
,
landete@umh.es
,
jalcaraz@umh.es
*Presenting author.
Keywords: Linear Ordering Problem, Bilevel Programming, Routing
Introduction
Routing, inside the category of location problems, can be interpreted and modelled as ranking
problems: the first locations to be visited are those with the best ranking scores and the
direction of the route indicates the position in the ranking. Therefore, the position that an
element occupies in a route or ranking usually depends on different criteria.
Considering a given route and focusing on one of the visited elements in the route that is not
in first place and that has budget to improve its position, in this work we model the problem
of deciding in which criteria it should improve in order to be visited as soon as possible. We
assume that the order of the route is calculated by solving the Linear Ordering Problem [1]
that makes use of the information in the different criteria. We propose a bilevel model in which
the first level improves the position of the selected element and the second level solves the
Linear Ordering Problem. We propose a genetic heuristic resolution algorithm and compare
the results with the exact resolution for instances where the exact can be solved. We analyze
the key properties of the problem and the optimization model.
Bibliography
[1]
J. Alcaraz, M. Landete and J. F. Monge. Rank aggregation: models and algorithms. Salhi
and Boylan (Eds):The Palgrave Handbook of Operations Research, 2022.
73
XIII International Workshop on Locational Analysis and Related Problems
Optimization approaches to scheduling working shifts for train
dispatchers
Ramón Piedra-de-la-Cuadra a,*, Francisco A. Ortega a
aUniversidad de Sevilla, rpiedra@us.es,riejos@us.es
*Presenting author.
Keywords: Railway dispatching, Shift scheduling, Integer Programming
Introduction
Scheduling work shifts involves the strategic assignment of tasks, the determination of their
locations, start and end times and, accordingly, the allocation of personnel. For railway
dispatchers, this responsibility is paramount as they plan, coordinate, and oversee train
movements within the rail network, ensuring safety and efficiency. The complexity of this
task is magnified by various constraints such as legal limitations on shift duration and on
the consecutive night shifts, in addition to other operational restrictions like the number of
areas a dispatcher can cover, and the need to avoid undesirable schedules. Traditionally, train
dispatcher schedules have been manually crafted, which is time-consuming, and it can be
prone to errors given the high number of constraints to be considered. To address this, an
optimization model is proposed to automate the scheduling process. By automating shift
scheduling using optimization models, railway companies can streamline operations, reduce
administrative burden, and improve overall efficiency. Moreover, by considering both legal
requirements and employee preferences, these automated systems can foster a better work-life
balance and enhance employee morale. In conclusion, the implementation of optimization
models offers a promising approach to address the complexities of scheduling work shifts for
train dispatchers while ensuring safety, efficiency, and the employee well-being.
Our proposed model considers various legal limitations, assignments of feasible areas, and
the quality of shifts to avoid undesirable configurations, even if permissible by law or union
agreements. Input data for the model includes geographical areas within the railway network,
feasible combinations of zones based on the existing infrastructure, and initial assignments
for dispatchers according to their preferences. Additionally, constraints such as task load
per dispatcher, shift duration limits, rest periods, and controllable areas per dispatcher are
incorporated. The objective is to generate schedules that minimize the number of dispatchers
while adhering to legal and operational constraints and optimizing employee satisfaction.
75
Granada, 4-6 September 2024 IWOLOCA 2024
Bibliography
[1]
Lapègue, T., Bellenguez-Morineau, O., and Prot, D. (2013). A constraint-based approach
for the shift design personnel task scheduling problem with equity. Computers & Operations
Research 40.10, pp. 2450–2465.
[2]
Hur, Y. et al. (2019).A stochastic optimization approach to shift scheduling with breaks
adjustments. Computers & Operations Research 107, pp. 127–139.
[3]
Lidén, T., Schmidt, C., and Zahir, R. (2023). Shift Scheduling for Train Dispatchers.
RailBelgrade 2023: the 10th International Conference on Railway Operations Modelling
and Analysis (ICROMA).
76
XIII International Workshop on Locational Analysis and Related Problems
The Hampered K-Median Problem with Neighbourhoods
Justo Puerto a, Carlos Valverde a,*
aInstituto de Matemáticas, Universidad de Sevilla, puerto@us.es,cvalverde@us.es
*Presenting author.
Keywords: Facility location, Continuous location, Barriers, Mixed integer Conic programming
Abstract
This paper deals with facility location problems in a continuous space with neighbours and
barriers. Each one of these two elements, neighbours and barriers, makes the problems
harder than their standard counterparts. Combining all together results in a new challenging
problem that, as far as we know, has not been addressed before, but has applications for
inspection and surveillance activities and the delivery industry assuming uniformly distributed
demand in some regions. Specifically, we analyse the
𝑘
-Median problem with neighbours and
polygonal barriers in two different situations. None of these problems can be seen as a simple
incremental contribution since in both cases the tools required to analyse and solve them go
beyond any standard overlapping of techniques used in the separated problems. As a first
building block, we deal with the problem assuming that the neighbourhoods are not visible
from one another and therefore there are no rectilinear paths that join two of them without
crossing barriers. Under this hypothesis, we derive a valid mixed-integer linear formulation.
Removing that hypothesis leads to a more general and realistic problem, but at the price of
making it more challenging. Adapting the elements of the first formulation, we also develop
another valid mixed-integer bilinear formulation. Both formulations rely on tools borrowed
from computational geometry that allow to handle polygonal barriers and neighbours that
are second-order cone (SOC) representable, which we preprocess and strengthen with valid
inequalities. These mathematical programming formulations are also instrumental to derive an
adapted matheuristic algorithm that provides good quality solutions for both problems in short
computing time. The paper also reports extensive computational experience, counting 2400
experiments, showing that our exact and heuristic approaches are useful: the exact approach
can solve optimally instances with up to 50 neighbourhoods and different number of barriers
within one hour of CPU time, whereas the matheuristic approach always returns good feasible
solutions in less than 300 seconds.
77
XIII International Workshop on Locational Analysis and Related Problems
Exact approaches for the Chinese Postman Problem with
load–dependent costs
Paula Segura a,*, Isaac Plana b, José María Sanchis a
aUniversitat Politècnica de València, psegmar@upvnet.upv.es,jmsanchis@mat.upv.es
bUniversitat de València, isaac.plana@uv.es
*Presenting author.
Keywords: arc routing problems, load-dependent costs, mathematical formulations, branch
and cut
Introduction
The Chinese Postman Problem (CPP) is a classical arc routing problem whose goal is to find
a minimum-cost tour on a connected undirected graph that traverses each edge at least once
([1]). In transportation systems, the level of emissions from a vehicle is influenced by factors
beyond the distance traveled, such as its load. Motivated from the desire to reduce pollution,
the Chinese Postman Problem with load–dependent costs (CPP-LC) was introduced in [2]
as an extension of the CPP in which the cost of traversing an edge depends on its length
and also on the weight of the vehicle at the moment the edge is traversed. In this talk, we
summarize the two mathematical programming formulations proposed in the literature for the
CPP-LC and present a new mixed-integer linear programming formulation for the problem,
proposing several families of valid inequalities to reinforce such a formulation. We design a new
branch-and-cut algorithm for the CPP-LC solution based on this formulation that incorporates
the separation of the valid inequalities proposed. Some computational results obtained with
our new exact procedure are compared with those already existing in the literature.
Bibliography
[1]
Laporte, G. The undirected Chinese postman problem. Arc Routing: Problems, Methods,
and Applications. MOS–SIAM Series Optimization, 20, 53-64, 2014.
[2]
Corberán, Á., Erdogan, G., Laporte, G., Plana, I., Sanchis, J.M. The Chinese Postman
Problem with Load–Dependent Costs. Transportation Science, 52(2), 370–385, 2018.
79
XIII International Workshop on Locational Analysis and Related Problems
New advances in hypergraph structure analysis
Francisco Temprano a,*, Stefano Benati b, Justo Puerto a
aUniversidad de Sevilla, ftgarcia@us.es,puerto@us.es
bUniversità degli Studi di Trento, stefano.benati@unitn.it
*Presenting author.
Keywords: Network location, Mathematical programming, Branch-and-Price
Abstract
This talk deals with the problem of locating subsets of nodes highly con nected over hypergraphs.
Due to the lack of universal consensus and de veloped theory in literature, we must come
up with a formal definition aim of the problem, in order to propose new optimization models
to solve it. A hypergraph is known to be a generalization of a graph that allows us to
represent a huge amount of real-life interactions between elements of a data sample that the
classic and original networks can not. Once we for mally define the problem, we develop a
large list of functions able to mea sure the goodness of the node set subdivision. Next, we
compare all these methods by using of mathematical programming and extensive computa
tional experiments. Thus, we can conclude which methods describe better the properties of
partition structures and which ones perform better, since multiple compact formulations and
column generation algorithms are de veloped to solve the partitioning hypergraph problem.
Finally, we have applied all methods to Eurobarometer data, showing the applicability of the
introduced methodology.
Bibliography
[1]
Diego Ponce, Justo Puerto, and Francisco Temprano. "Mixed integer linear programming
formulations and column generation algorithms for the minimum normalized cuts problem
on networks." (2023) Available at SSRN 4354039.
[2]
Zhou, Dengyong, Jiayuan Huang, and Bernhard Schölkopf. "Learning with hypergraphs:
Clustering, classification, and embedding." Advances in neural information processing
systems 19 (2006).
81
XIII International Workshop on Locational Analysis and Related Problems
Cross-docking platforms design and management under uncer-
tainty
Aitziber Unzueta a,* , Laureano F. Escudero b, María Araceli Garín c
a
Applied Mathematics Department, University of the Basque Country (UPV/EHU), Bilbao
(Bizkaia), Spain, aitziber.unzueta@ehu.eus
b
Statistics and Operations Research Area, University King Juan Carlos (URJC), Móstoles
(Madrid), Spain, laureano.escudero@urjc.es
c
Quantitative Methods Department, University of the Basque Country (UPV/EHU), Bilbao
(Bizkaia), Spain, mariaaraceli.garin@ehu.eus
*Presenting author.
Keywords: Cross-dock door design, Two-stage stochastic quadratic combinatorial optimiza-
tion, Linearized mixed-integer programming, Constructive matheuristic.
The Cross-dock Door Design Problem (CDDP) consists of deciding on the number and
capacity of inbound and outbound doors, minimizing the construction and exploitation cost
of the infrastructure. The uncertainty, realized in scenarios, lies in the occurrence of these
nodes, the number and cost of the pallets, and the capacity’s disruption of the doors. The
CDDP is represented using a stochastic two-stage binary quadratic model (BQM). Given the
difficulty of solving this combinatorial problem, a mathematically equivalent MILP formulation
is introduced. Searching an optimal solution of the whole MILP problem is still impractical,
thus, a scenario cluster decomposition-based matheuristic algorithm is presented and a broad
computational study to validate the proposal is carried out.
Introduction
Given a network with a set of supplying (i.e. origin) nodes for different product types and a
set of receiving (i.e. destination) nodes for these products, usually a cross-dock entity may
serve as a consolidation point. The origin nodes can deliver the material at the cross-dock so
that, after being classified by type and destination, it can be transported to the destination
nodes. A cross-dock infrastructure has a number of inbound doors, which are called strip
ones, and a number of outbound doors, which are called stack ones, and each of them has a
capacity for pallet handling during a given time period.
Two main types of optimization problems arise in cross-dock dealing. One of them is the
deterministic Cross-dock Door Assignment Problem (CDAP), see [3] and [2] for further
information about solving methodologies. The other one is the stochastic CDDP, the subject
83
Granada, 4-6 September 2024 IWOLOCA 2024
of this work. Comprehensive reviews on uncertainty have been recently published in [1].
In this work we introduce a two-stage stochastic BQP model for cross-dock door design
planning as an extension of the deterministic CDAP to deal with the uncertainty. The first
stage is devoted to the strategic decisions (i.e. the number of strip and stack doors and
related nominal capacities), before the uncertainties related to the set of origin and destination
nodes and the disruption of the doors capacity are unveiled. The second stage is devoted to
the operational decisions (i.e. the assignment of doors to origin and destination nodes in the
scenarios). The objective function to minimize is composed of the binary linear cross-dock door
infrastructure construction cost and the binary quadratic cost function related to the CDAP
scenario node-to-door assignment. We use the Reformulation Linearization Technique to
develop a Linearized mixed Integer Programming problem (LIP) as a equivalent reformulation
of the quadratic one. We introduce two options for a scenario Cluster Decomposition (CD)
of model LIP, where the special structure of the problem is benefitted from. The first option
decomposes model LIP into scenario clusters for the first stage variables. The second option
additionally does this for the strip and stack door problems. Finally, we develop a matheuristic
algorithm, based on feasible first stage solutions of the CD model, to obtain feasible solutions
for the stochastic model. It considers a linear search approach for large-scale instances to
exploit the scenario structure of the second stage submodels. It has been computationally
proved that the proposed scheme provides the same or a better incumbent solution value than
the plain use of CPLEX and it requires a much smaller wall time.
Bibliography
[1]
Ardakani, A., and Fei J. A Systematic Literature Review on Uncertainties in Cross-docking
Operations. Modern Supply Chain Research and Applications, 2020.
[2]
Escudero, L. F., Garín, M.A, and Unzueta, A. On Solving the Cross-dock Door Assignment
Problem. International Journal of Production Research, 2024.
[3]
Guignard, M. Strong RLT1 Bounds from Decomposable Lagrangean Relaxation for Some
Quadratic 0-1 Optimization Problems with Linear Constraints. Annals of Operations
Research 286: 173–200, 2020.
84
Author Index
A
Albareda, María
Universitat Politécnica de Catalunya, Spain, maria.albareda@upc.edu ........... 31
Alcaraz Soria, Javier
Universidad Miguel Hernandez de Elche, Spain, jalcaraz@umh.es ................ 73
Anitescu, Mihai
Argonne National Laboratory, USA, anitescu@mcs.anl.gov ..................... 25
B
Büsing, Christina
RWTH Aachen University, Germany, buesing@combi.rwth-aachen.de ........... 49
Baldassarre, Silvia
University of Naples Federico II, Italy, silvia.baldassarre@unina.it ........... 33
Baldomero-Naranjo, Marta
Universidad de Cádiz, Spain, marta.baldomero@uca.es . . . . . . . . . . . . . . . . . . 35, 37, 39
Benati, Stefano
Università degli Studi di Trento, Italy, stefano.benati@unitn.it ................ 81
Bessac, Julie
Argonne National Laboratory, USA, jbessac@anl.gov ........................... 25
Billhardt, Holger
Universidad Rey Juan Carlos, Spain, holger.billhardt@urjc.es ................ 71
Blanco, Víctor
Universidad de Granada, Spain, vblanco@ugr.es . . . . . . . . . . . . . . . . . 31, 41, 43, 45, 47
Bruno, Giuseppe
University of Naples Federico II, Italy, giuseppe.bruno@unina.it ................ 33
C
Camacho-Vallejo, José-Fernando
85
Granada, 4-6 September 2024 IWOLOCA 2024
Tecnologico de Monterrey, Mexico, fernando.camacho@tec.mx .................. 61
Canca, David
Universidad de Sevilla, Spain, dco@us.es ....................................... 51
Cattaruzza, Diego
Centrale Lille, France, name1@institution1.org ............................... 53
Cavola, Manuel
Pegaso University, Italy, manuel.cavola@pegaso.it ............................. 33
D
Domínguez, Concepción
University of Murcia, Spain, concepcion.dominguez@um.es ..................... 55
E
Escudero, Laureano F.
Universidad Rey Juan Carlos, Spain, laureano.escudero@urjc.es . . . . . . . . 57, 67, 83
Espejo, Inmaculada
Universidad de Cádiz, Spain, inmaculada.espejo@uca.es ....................... 59
F
Fernández-Gutiérrez, Juan Pablo
Universidad de Medellín, Colombia, jpfernandez@udemedellin.edu.co . . . . . . . . . . 61
Ferris, Michael
University of Wisconsin-Madison, USA, ferris@cs.wisc.edu .................... 25
Filippi, Carlo
University of Brescia, Italy, carlo.filippi@unibs.it ........................... 63
G
Gázquez, Ricardo
Universidad Carlos III de Madrid, Spain, ricardo.gazquez@uc3m.es .............. 35
Gagrani, Praful
University of Wisconsin-Madison, India, praful.gagrani@gmail.com ............. 41
Garín, María Araceli
Universidad del País Vasco, Spain, mariaaraceli.garin@ehu.eus . . . . . . . . 57, 67, 83
86
XIII International Workshop on Locational Analysis and Related Problems
González, Gabriel
Universidad de Granada, Spain, ggdominguez@ugr.es ........................... 41
Guastaroba, Gianfranco
University of Brescia, Italy, gianfranco.guastaroba@unibs.it .................. 63
H
Hinojosa, Yolanda
Universidad de Sevilla, Spain, yhinojos@us.es .................................. 31
J
Japón, Alberto
Universidad de Sevilla, Spain, ajapon1@us.es ................................... 43
K
Kalcsics, Jörg
The University of Edinburgh, United Kingdom, joerg.kalcsics@ed.ac.uk . . . . . . . 37
Krock, Mitchell
Argonne National Laboratory, USA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
L
López-Sánchez, Aitor
Universidad Rey Juan Carlos, Spain, aitor.lopez@urjc.es ...................... 71
Labbé, Martine
Université Libre de Bruxelles, Belgium, martine.labbe@ulb.be .................. 53
Landete Ruiz, Mercedes
Universidad Miguel Hernández de Elche, Spain, landete@umh.es . . . . . . . . . . . . . . 69, 73
Leitner, Markus
Vrije Universiteit Amsterdam, The Netherlands, m.leitner@vu.nl ................ 49
Luedtke, James
University of Wisconsin-Madison, USA, jim.luedtke@wisc.edu .................. 25
Lujak, Marin
Universidad Rey Juan Carlos, Spain, marin.lujak@urjc.es ...................... 71
87
Granada, 4-6 September 2024 IWOLOCA 2024
M
Mancuso, Andrea
University of Naples Federico II, Italy, andrea.mancuso@unina.it ................ 39
Marín, Alfredo
Universidad de Murcia, Spain, amarin@um.es ................................ 23, 59
Martínez-Antón, Miguel
Universidad de Granada, Spain, mmanton@ugr.com ............................... 45
Masone, Adriano
University of Naples Federico II, Italy, adriano.masone@unina.it ................ 39
Melero, Inigo Martin
Universidad Miguel Hernandez de Elche, Spain, inigo.martin@goumh.umh.es . . . . . 73
Muñoz-Ocaña, Juan Manuel
Universidad de Cádiz, Spain, juanmanuel.munoz@uca.es .................... 59, 69
O
Ortega, Francisco A.
Universidad de Sevilla, Spain, riejos@us.es .................................... 75
P
Páez, Raúl
Universidad de Cádiz, Spain, raul.paez@uca.es ................................ 59
Petris, Matteo
INRIA, France, matteo.petris@inria.fr ...................................... 53
Piedra-de-la-Cuadra, Ramón
Universidad de Sevilla, Spain, rpiedra@us.es ................................... 75
Pipicelli, Eduardo
University of Naples Federico II, Italy, eduardo.pipicelli@unina.it ............ 33
Plana, Isaac
Universitat de València, Spain, isaac.plana@uv.es ............................. 79
Pozo, Miguel A.
Universidad de Sevilla, Spain, pozo@us.es ...................................... 47
Puerto, Justo
Universidad de Sevilla, Spain, puerto@us.es . . . . . . . . . . . . . . . . . . . . . . . . . 43, 47, 77, 81
88
XIII International Workshop on Locational Analysis and Related Problems
R
Roald, Line
University of Wisconsin-Madison, USA, roald@wisc.edu ........................ 25
Rodríguez-Chía, Antonio M.
Universidad de Cádiz, Spain, antonio.rodriguezchia@uca.es . . . . 35, 37, 39, 59, 69
Roland, Marius
Polytechnique Montreal, Canada, mmmroland@gmail.com ........................ 53
Rossmann, Ramsey
University of Wisconsin-Madison, USA, rossmann2@wisc.edu .................... 25
S
Salazar-González, Juan-José
University of La Laguna, Spain, jjsalaza@ull.es .............................. 63
Saldanha-da-Gama, Francisco
Sheffield University, United Kingdom,
francisco.saldanha-da-gama@sheffield.ac.uk ................................. 69
Salman, Sibel
Koç University, Turkey, ssalman@ku.edu.tr .................................... 27
Sanchis, José María
Universitat Politècnica de València, Spain, jmsanchis@mat.upv.es ............... 79
Schmidt, Martin
Trier University, Germany, martin.schmidt@uni-trier.de ...................... 53
Segura, Paula
Universitat Politècnica de València, Spain, psegmar@upvnet.upv.es .............. 79
Semet, Frederic
Centrale Lille, France, frederic.semet@centralelille.fr ..................... 71
T
Temprano, Francisco
Universidad de Sevilla, Spain, ftgarcia@us.es .................................. 81
Torrejón, Alberto
Universidad de Sevilla, Spain, atorrejon@us.es ................................. 47
89
Granada, 4-6 September 2024 IWOLOCA 2024
U
Unzueta, Aitziber
Universidad del País Vasco, Spain, aitziber.unzueta@ehu.eus . . . . . . . . . . 57, 67, 83
V
Valverde, Carlos
Universidad de Sevilla, Spain, cvalverde@us.es ................................. 77
Villegas, Juan G.
Universidad de Antioquia, Colombia, juan.villegas@udea.edu.co ............... 61
W
Wang, Xingrong
School of Systems Science, Beijing , China 20114009@bjtu.edu.cn .............. 51
Wrede, Sophia
RWTH Aachen University, Germany, wrede@combi.rwth-aachen.de ............. 49
Z
Zhang, Peter
Carnegie Mellon University, United States of America, pyzhang@cmu.edu . . . . . . . . . . 43
90
Useful Information
Venues
Talks will be held at two buildings: the Instituto de Matemáticas de Granada (IMAG) and
El Carmen de la Victoria.
•
The address of the IMAG is at Ventanilla nº11 de Granada. The way of arriving at the
institute is by Rector López-Argüeta Street. Follow the red path in the picture below.
Access to IMAG.
•
The address of El Carmen de la Victoria is at Encrucijada, 13. It is 30 minutes far
away from the IMAG.
Press the following link for directions in Google maps from IMAG to El Carmen de la
Victoria: https://maps.app.goo.gl/ufN4UTY9kMEmQmKZA
Our venues are at 1.8 km from each other. Commuting between them will take 28 min walking
distance. Uber/Taxi is available in Granada, as well as urban bus (take one bus from Triunfo
to Plaza Nueva and another from Plaza Nueva to Paseo de los Tristes).
91
Granada, 4-6 September 2024 IWOLOCA 2024
Wi-Fi
Wi-Fi will be available during the conference using eduroam network.
Social Activities
★Wednesday, September 4, 2024: Welcome Reception
On Wednesday, it will take place the welcome reception at 9 pm at BHeaven, Calle Acera
del Darro, 62, at the rooftop of Hotel Barceló Carmen Granada.
★Thursday, September 5, 2024: Guided Tour
On Thursday, we will have a walk through Albayzin, the picturesque neighborhood, focusing
on the viewpoints (miradores). The tour will finish with a gastronomic tour to taste the
specialty of Granada: the tapas. The departure for this event will be at 6 pm, from El Carmen
de la Victoria.
★Friday, September 6, 2024: Gala Dinner
On Friday, our Gala dinner will be held at 9 pm at the gardens of El Carmen de la Victoria.
Interactive map
All locations related to the IWOLOCA 2024 conference can be found on this map, as well as
points of interest in the city of Granada:
https://bit.ly/IWOLOCA24_spots
92
XIII International Workshop on Locational Analysis and Related Problems
93
Partner Institutions and Sponsors
The IWOLOCA is very grateful to the following sponsors.
XIII IWOLOCA 2024 has been partially supported by Grant RED2022-134149-T funded by
MICIU/AEI /10.13039/501100011033 and IMAG-Maria de Maeztu grant CEX2020-001105-M
/AEI /10.13039/501100011033
95
