IO 2017

Book of Abstracts

June 28-30, 2017

Escola Superior de Ciências Empresariais de Valença

Instituto Politécnico de Viana do Castelo

Sponsors and Supporters

We would like to thank our sponsor and supporters for their contribution to the suc-cess of the conference.

Sponsors

Supporters

2 IO 2017

Contents

Opening Notes iii

Aims and Scope v

Organizing Committee vii

Program Committee ix

General Information 1

Social Program 3

Maps 5

Program Overview 9

Plenary Sessions 15

Sessions 19

Operação Nariz Vermelho 67

Tourism 69

EstudIO 71

Index 83

IO 2017 i

Opening Notes

Welcome to IO 2017!

On behalf of the organization committee, we welcome you to the 18th APDIO congress.

This year, the event will take place at Escola Superior de Ciências Empresariais deValença, Instituto Politécnico de Viana do Castelo. We will have the opportunity to enjoythe beautiful new facilities of this school and visit Valença, a fortified town located on ahill looking over the left bank of the Minho, opposite the Galician town of Tui, on the mainroad that connects Santiago de Compostela to Porto. The old district of Valença is a veryunusual twinned area, comprising two Vauban-style strongholds joined by an old steelbridge. The international bridge Tui-Valença is one of the most notable works of D. PelayoMancebo Y Agreda, a famous Spanish engineer.

We would like to thank the APDIO directive committee for entrusting us with part ofthe organization of this event and we sincerely hope that our effort will contribute to itssuccess.

In addition, we would also like to thank all the participants that chose to attend thiscongress of Operational Research. We will do our best to provide all you need to fullyenjoy this event. Aside from the expected presentations, there will be spaces to share andexchange ideas.

The social program includes a small visit to the Valença Fortress, a Nacional Monument, awelcome drink on the first day and a dinner on the second day. All was arranged in orderto show you a peek of the close surroundings and allow you to re-establish the energy tocontinue work.

Please don’t hesitate to contact any member of the organization committee for anythingyou may need during the event.

IO 2017 iii

Aims and Scope

IO2017 aims to bring together researchers from different scientific areas and with distinctbackgrounds, but with common interests in operations research. It should be an impor-tant forum of discussion and an environment favorable to the exchange of ideas.

The EstudIO is an important part of the congress since it encourage young students topresent their works in an academic environment and helps them in their first steps in theresearch world.

It is the eighteenth edition of a series of international conferences in operations researchorganized in Portugal under the auspices of APDIO (the Portuguese Operational ResearchSociety). As in previous editions, APDIO will deliver some awards, to PhD theses, to thebest papers in Process Systems Engineering of APDIO members, to the best papers of AP-DIO members, and to the most voted work of EstudIO.

This year’s event theme is Operations Research Pro Bono and some relevance is dedicatedto a chosen nonprofit organization, Operação Nariz Vermelho, that tries to take some com-fort and joy to hospitalized children.

IO 2017 v

Organizing Committee

Lia Oliveira Instituto Politécnico de Viana do Castelo ChairEscola Superior de Ciências Empresariais

Sofia Rodrigues Instituto Politécnico de Viana do CasteloEscola Superior de Ciências Empresariais

António Amaral Instituto Politécnico de Viana do CasteloEscola Superior de Ciências Empresariais

Filipe Carvalho Instituto Politécnico de Viana do CasteloEscola Superior de Ciências Empresariais

Luís Barreto Instituto Politécnico de Viana do CasteloEscola Superior de Ciências Empresariais

Sandra Silva Instituto Politécnico de Viana do CasteloEscola Superior Agrária

IO 2017 vii

Program CommitteeA. Ismael F. Vaz, UM, EE Chair

Agostinho Agra, UA, DM José Fernando Gonçalves, UP, FEconAna Cristina Amaro, IPC, ISCAC | CBS José Fernando Oliveira, UP, FEAna Isabel Pereira, IPB, ESTG José Manuel Valério de Carvalho, UM, EEAna Luisa Custódio, UNL, FCT Lia Oliveira, IPVC, ESCEAna Paula Barbosa Póvoa, ULisboa, IST Luís Miguel Cândido Dias, UC, FEconAna Paula Teixeira, UTAD, Luís Nunes Vicente, UC, FCTArmando B. Mendes, UA, FCTs Margarida Vaz Pato, ULisboa, ISEGBernardo Almada-Lobo, INESC TEC e UP, FE Maria Antónia Carravilla, UP, FECândida Mourão, ULisboa, CMAF-CIO e ISEG Maria Eugénia Captivo, ULisboa, FCCarlos Henggeler Antunes, UC, DEEC Maria João Alves, UC, FEconCesaltina Pires, UE, DG Maria Sameiro Carvalho, UM, EEClara Bento Vaz, IPB, ESTIG Marília Pires, UAlg, FCTCristina Lopes, IPP, ISCAP Mónica Oliveira, ULisboa, ISTDomingos Moreira Cardoso, UA, DM Paulo Sérgio Abreu Freitas, UMad, FCEEFernando José Vicente Freire, AM Pedro Oliveira, UP, ICBASFilipe Alvelos, UM, EE Silvério Simões Rosa, IT e UBIIsabel Gomes, UNL, FCT Sofia Rodrigues, IPVC, ESCEJoão Miranda, IPPort, CERENA, IST Susana Relvas, ULisboa, ISTJoão Paulo Costa, UC, FEcon Tatiana Tchemisova Cordeiro, UA, DMJoão Telhada, ULisboa, FC Teresa Cardoso Grilo, IUL, ISCTE BSJorge Orestes Cerdeira, UNL, DM Tomaz Dentinho, UA, DCA

IO 2017 ix

General Information

LanguageThe conference official languages are Portuguese and English. No simultaneous translation will beprovided.

Conference VenueThe conference is taking place at the Escola Superior de Ciências Empresariais de Valença do Insti-tuto Politécnico de Viana do Castelo. A map can be found on page 6, indicating the location wheresessions, coffee breaks, and lunch are taking place. Signs disposed in the buildings will guide theparticipants to the sessions rooms.

Registration DeskThe registration desk will be located at the entrance near the main Auditorium. The registration deskwill be opened on Wednesday 28th, from 12:00 to 18:30, and during the remaining days, from 09:00to 13:00.

Internet AccessFree wireless access is available through the campus (network name: ESCE-WLAN, without pass-word - Warning: this is an unsecured wireless network).

Guidelines for Session ChairsIn the Session Chairs Index on page 85 you can find the codes of all the sessions chaired by a givenparticipant. The session location is given in the Abstracts section of the Conference Book. The role ofthe chair is to coordinate and ensure the smooth execution of the session. As a session chair pleasemake sure to:

• Contact the speakers before the session, to verify who is presenting and to preempt any tech-nical problems.

• Ensure that the session begins and ends on time. All sessions lasts 20 minutes. Time forquestions is included. Sessions are organized with 4 or 5 speakers. Open slots may exist in theend of the session.

• Introduce the speaker and the title of each talk.

IO 2017 1

GENERAL INFORMATION

• Ensure that presentations respect the program order, to allow participants to jump betweensessions. If a speaker cancels or does not attend, the session schedules should be respected,rather than shifting every talk backwards.

• When appropriated inform the speaker of the remaining time till the end of the presentation.

• Ask for questions and thank the speaker.

VERY IMPORTANT: We ask session chairs to notify us about any last minute changes or cancel-lations. These changes will be posted outside the meeting rooms.

Guidelines for SpeakersIn the Session Speakers Index on page 83, next to the speaker name, you can find the code of thesession in which a given speaker participates and the corresponding page number. All session roomswill be equipped with laptops or desktop computers and overhead projectors. You may use your ownlaptop to ensure that your presentation use the right version of the software and fonts installed, sothat it looks like what you have planned and designed. Please follow these guidelines to ensure asuccessful presentation.

• If you bring your own laptop to your session, bring along the power supply cable. You mayneed an adapter to connect your computer to the local voltage (220) and wall plug type.

• If your laptop is a Mac, bring the required adapter for the external video output.

• Arrive at your session at least 10 minutes before it begins. All presenters in a session shouldset up and test the connection to the projector before the session begins.

• We encourage speakers to put their presentations on a Universal Serial Bus data stick (USBpen) as a backup.

• Time your presentation to fit the allotted time (20 minutes), including time for questions andaudience participation.

• Feel free to bring along copies of your paper to distribute or to provide a handout with relatedinformation. No proceedings with complete papers are to be delivered for this meeting. Toobtain complete copies of any papers abstracted in the program, please contact the authorsdirectly at the address supplied with each abstract.

• One or more session assistants (student volunteers) will be available at each room. You canaddress the session assistant for any request or help regarding problems related to audiovisualequipment.

LunchesLunches will be served in the restaurant facilities. You are kindly required to wear your badge.

Coffee-breaksCoffee, tea, juices, and bottled water will be available during the conference coffee-breaks.

TransportationTransportation to the conference site is available according to the program on page 10. The BUS de-partures from Hotel Minho and goes by Hotel Lara, arriving at the Escola Superior de Ciências Em-presariais/ Business Sciences School. For more information please contact an organization member.

Valença is a small city, where almost all visiting sites are in a walking distance.

2 IO 2017

Social Program

• Wednesday 28th June, from 19:00 to 20:00: Tour to the Valença Fortress.

The tour will allow to know some of the most amazing places and stories of the ValençaFortress. The visit will be organized in three groups with three guides. Two groups will beguided in Portuguese language and the other in English language.

• Wednesday 28th June, from 20:00 to 22:30: Welcome drink

This special welcome drink will be served in the Pousada de S. Teoónio, a magnificent placeinside the Fortress.

• Thursday 29th June, from 20:30 to 24:00: Conference dinner

Conference dinner will be served in Quinta do Amorinho.

IO 2017 3

Maps

CampusEscola Superior de Ciências Empresariais de ValençaInstituto Politécnico de Viana do Castelo Av. Pinto da Mota4930-600 Valença Telef: 258 809 679Telm.: 937 300 026 | Fax: 251 800 841 Site: www.esce.ipvc.ptE-mail: geral@esce.ipvc.pt

(GPS: [latitude: 42.03163o N - longitude: -8.63296o E])

IO 2017 5

MAPS

Conference RoomsSessions take place in seven rooms: main Auditorium, Room 1, Room 2, Room 3, Room 4, Room 5,and Room 6.

6 IO 2017

MAPS

Directions to the Conference SiteFrom Hotel Minho to the conference site

From Hotel Lara to the conference site (walking)

IO 2017 7

MAPS

From Pousada de S. Teotónio to the conference site (walking)

From the conference site to the conference dinner site (Quinta do Amorinho)

8 IO 2017

Program Overview

29

/jun

30

/jun

09:00

Shu

ttle Ho

tel Min

ho

- Ho

tel Lara - ESCE

09:00

Shu

ttle Ho

tel Min

ho

- Ho

tel Lara- ESCE

10:50 - 1

1:20C

offee

Break

10:50

- 11

:20C

offee

Break

12:00

12:20

- 13

:00A

wa

rds C

eremo

ny a

nd

Clo

sing

Session

13:00

Shu

ttle Ho

tel Min

ho

- Ho

tel Lara - ESCE

14:00

16:00

- 16

:20

16:20

- 16

:50N

ote (1):

16:50 - 1

7:20C

offee

Break

18:35

19:05

Shu

ttle ESCE - H

otel Lara - H

otel M

inh

o

20:00

Shu

ttle Ho

tel Min

ho

- Ho

tel Lara - "Qu

inta d

o A

mo

rinh

o"

22:30

00:00

Shu

ttle "Qu

inta d

o A

mo

rinh

o" - H

otel Lara - H

otel M

inh

o

10:45

(Arrival@

12:00

)

Shu

ttle Braga - V

alença (1

)

14:3

0

(Arrival@

15:45

)

17:20 - 1

9:00

20:30 - 0

0:00

14:00

- 15

:00

15:00

- 16

:00

16:50

- 18

:30

19:00

- 20

:00

20:00

- 22

:30

Shu

ttle ESCE - M

uralh

as

Tou

r: Mu

ralhas d

e Valen

ça

Welco

me: "A

lvarinh

o d

e Ho

nra"

Shu

ttle Valen

ça - Ho

tel Min

ho

Plen

ary Session

III

Ricard

o V

argasP

arallel Session

s (five talks per sessio

n)

AP

DIO

Gen

eral A

ssemb

ly

Co

nferen

ce Din

ner

Lun

ch

28

/jun

Op

enin

g Sessio

n

Plen

ary Sessio

n I

An

a Pau

la Pó

voa

Parallel Sessio

ns (five talks p

er session

)

Registratio

n

Shu

ttle Braga - V

alença

Parallel Sessio

ns (fo

ur talks p

er session

)P

arallel Session

s (fou

r talks per sessio

n)

Lun

ch

IO P

ro B

on

o: "R

ir é o M

elho

r Rem

édio

?"

Co

ffee B

reak

09:30

- 10

:5009

:30 - 10:50

15:30 - 1

6:50

A lu

ggage storage ro

om

will b

e available d

urin

g

the m

orn

ing o

f Jun

e 30.

11:20

- 12

:20

13:00

- 14

:30

11:20 - 1

3:00

13:00 - 1

4:30

14:30 - 1

5:30

Parallel Sessio

ns (fo

ur talks p

er session

)

Plen

ary Session

II

Ru

th K

aufm

an

IO 2017 9

PROGRAM OVERVIEW

Wednesday, June 28, 16:00-16:20

Wed.0.0 OR Pro Bono, Session, Auditorium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19C. Hiolanda Esteves "Rir é o Melhor Remédio?": Estudo do Impacto da Intervenção dos Doutores

Palhaços da Operação Nariz Vermelho

Wednesday, June 28, 16:50-18:30

Wed.1.1 EstudIO, Session, Room 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Ana Raquel Aguiar Planning home health care services - a routing and scheduling problemDaniel Bandeira Impressão tridimensional bimaterial por processo aditivoRaphael Cordeiro Sistema de apoio à decisão para elaboração de mapas de exames no ensino supe-

riorJoão Gonçalves Optimal control strategies for an advertisement viral diffusionJuliana Marques Impacto da uniformização e centralização de consumos em contexto hospitalar

Wed.1.2 DEA and Performance Analysis, Session, Room 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Renata Oliveira The evaluation of the social performance of mining firmsRui Cunha Marques Malmquist and Hicks-Moorsteen productivity indexes for clusters performance

evaluationRui Marques Análise analítica da eficiência de rega usando modelos determinísticos e estocásticosAntónio Fernandes Efficiency and Capital Structure in Portuguese SMEsAndré S. Veloso Determinants of nursing homes performance: The case of Portuguese Santas Casas

da MisericórdiaWed.1.3 Production and Inventory Management, Session, Room 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Serhat Saylam A spreadsheet based model that estimates the impact of reduced distribution time oninventory investment savings

Maria Pires Solving the backroom design problemJorge Mendes Review policies for online retail inventory management considering customer order

time windowsJoão Correia Pinto A prescriptive analytics method applied to a maintenance and manpower

scheduling problemPaula Alves Desenvolvimento de uma metodologia para redução do volume de obsoletos numa em-

presa do sector automóvelWed.1.4 Data Analysis and Artificial Inteligence, Session, Room 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Alvaro Neuenfeldt Júnior The two-dimensional strip packing problem: what matters?Isabel Cristina Lopes The reverse logistics of unsold medications in pharmacies in Campania, ItalyJosé Oliveira Optimization of short-term storage of containers in empty container terminalsGonçalo Figueira Automatic learning of planning rules for operations management problemsHelena Alvelos Melhoria do processo de corte a laser numa empresa metalomecânica

Wed.1.6 Logistics and Supply Chain, Session, Room 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26Bruna Mota Green supply chain design and planning: the importance of decision integration in

optimization modelsPedro Amorim Leveraging logistics flows to improve the sludge management process of wastewater

treatment plantsNuno Falcão e Cunha Supply chain modelling and supplier selection in the aerospace industryAna Cristina Amaro Pharmaceutical Supply Chain: a comprehensive optimal planning for

medicine shortageAna Paula Barbosa Póvoa Design and planning of sustainable supply chains - the case study of a

tissue paper business

10 IO 2017

PROGRAM OVERVIEW

Thursday, June 29, 9:30-10:50

Thu.1.1 Other Models, Methods, and Algorithms, Session, Room 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Maria Antónia Carravilla Understanding complexity in a practical combinatorial problem using

Mathematical Programming and Constraint ProgrammingManuel Vieira A formulation for the multi-row layout problemMaria João Alves Ilustração de diferentes tipos de soluções em problemas de programação em dois

níveis semivetorialThu.1.2 Multicriteria I, Session, Room 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Manuel Matos Abordagem multicritério ao problema de expansão de um sistema produtor elétricoDalila B. M. Martins Fontes Evaluating suppliers in the olive oil sector using AHPAna Sara Costa The quality of service: An overall performance assessment for water utilitiesLuis C. Dias Uma abordagem de avaliação multicritério para classificação e ordenação de alterna-

tivas conjugando Delphi e ELECTRE TRIThu.1.3 Multiobjective, Session, Room 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Saman Aminbakhsh An efficient heuristic method for Pareto oriented optimization of timecosttradeoff problem for construction projects

João Soares Use of analytic hierarchy process (AHP) to support the decision-making about destina-tion of a batch of defective products with alternatives of rework and discard

Helga Pereira Multiobjective irrigation model. The Alqueva river basin.Sandra Silva A logística humanitária e a logística empresarial: relações e desafios.

Thu.1.4 Health and Life Science I, Session, Room 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32Diogo Cunha Ferreira Economies of scope in health sector: the case of Portuguese hospitalsSofia Cruz-Gomes What future for healthcare expenditures? A time series analysisLiliana Freitas A novel participatory approach to scenario building: application to the evolution of

population health inequalities in EuropeFilipe Alvelos A priori crossmatching in kidney exchange programs

Thu.1.5 Discrete Optimization, Graphs, and Geometry I, Session, Room 5 . . . . . . . . . . . . . . . . . . . . . . . . . 34Margarida Carvalho Existence of Nash equilibria on integer programming gamesRicardo Soares Planning wood chips delivery to bioenergy centralsWilliam Hitoshi Tsunoda Meira A decomposition approach for the long-term scheduling of a single-

source multiproduct pipeline networkDomingos Cardoso Grafos com número de estabilidade quadrático convexo: principais resultados

e problemas em abertoThu.1.6 Location, Logistics, Transportation, and Traffic I, Session, Room 6. . . . . . . . . . . . . . . . . . . . . . . . .35

Arsham Atashi Khoei Energy minimizing p-connected covering location problemElsa Silva An intercontinental replenishment problem: A hybrid approachJosé Correia Otimização de rotas para abastecimento de linhas de montagemAna Moura Otimização da gestão de agendamento de prestação de cuidados de saúde

Thursday, June 29, 11:20-13:00

Thu.2.1 Production Management and Supply Chain, Session, Room 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Fábio Neves-Moreira The time window assignment vehicle routing problem with product depen-

dent deliveriesLarissa Oliveira A matheuristic for the irregular strip packing problemAna Raquel Xambre Postponement: definição do ponto de desacoplamento com recurso a um pro-

cedimento para a formação de famílias de produtos

IO 2017 11

PROGRAM OVERVIEW

Thu.2.2 Multicriteria II, Session, Room 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Marcio Basilio Mapping the use of multicriteria methods applied to public securityMax Oliveira Métodos de decisão multicritérios: um estudo empírico aplicado à segurança públicaAna Vieira Shaping Delphi processes to inform multicriteria decision analysis in the healthcare con-

textMonica Oliveira Overcoming common critical mistakes in building composite indices through the

MACBETH socio-technical approachThu.2.3 Energy, Environment, Natural Resources, and Weather I, Session, Room 3 . . . . . . . . . . . . . . . . 40

Andreia Santos OR methods applied to forest-based supply chains considering economic, environ-mental and social perspectives - A systematic review

Leão José Fernandes Downstream petroleum supply chains’ design and planning - contributionsand roadmap

Alexandra Marques A collaborative platform to enhance sustainability in forest-based supplychains

Amilcar Arantes Towards sustainability in the construction and demolition waste industry: a casestudy on application of lean methodologyThu.2.4 Health and Life Science II, Session, Room 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Susana Rafaela Martins Warfarin dose optimization with genetic algorithmsJoana Faria Planning health workforce training in the detection and prevention of excessive alcohol

consumption: an optimization-based approachTeresa Cardoso-Grilo Planning the delivery of long-term care: Improving location-allocation multi-

objective stochastic models through scenario reduction techniquesMarta Monteiro Planning healthcare workforce training: a multi-objective mathematical program-

ming modelMaria Eugénia Captivo Local search heuristics for a surgical case assignment problem

Thu.2.5 Nonlinear Optimization, Session, Room 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

M. Fernanda P. Costa A penalty approach for solving nonsmooth and nonconvex MINLP problemsA.I.F. Vaz A branch and bound method for convex semi-infinite programmingTatiana Tchemisova A generator of nonregular semidefinite programming problemsSérgio Pereira Can optimization help additive manufacturing?Ana Luisa Custodio An approach based on nonmonotone directional direct search to noisy opti-

mizationThu.2.6 Location, Logistics, Transportation, and Traffic II, Session, Room 6 . . . . . . . . . . . . . . . . . . . . . . . . 46

Beatriz Brito Oliveira A dynamic programming approach for integrating dynamic pricing andcapacity decisions in a rental context

António Ramos Cargo stability in the container loading problem - State-of-the-art and futureresearch directions

Leonor S. Pinto Rotas para recolha de valores nos arcosCarolina Soares de Morais Waste collection planning based on real-time informationMaria Isabel Gomes A biased-randomized heuristic for the home healthcare routing problem

Thursday, June 29, 15:30-16:50

Thu.3.1 Scheduling and Sequencing I, Session, Room 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Afshin Mehrsai Decentralized sequencing in real-time in job scheduling problemHendrik Vermuyten Optimizing staff scheduling: a case at INEMSara Martins Heurística de escalonamento para um laboratório de análises de vinhosRui Borges Lopes Heuristic approaches for a flexible job-shop scheduling problem

12 IO 2017

PROGRAM OVERVIEW

Thu.3.2 Multicriteria III, Session, Room 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Luis Dias Optimising costs and risks in asset managementJéssica de Freitas Investment portfolio formation via multicriteria decision aid: A brazilian stock

market studyAlberto Martinho Location-routing problems: Sectorization and multicriteria

Thu.3.3 Energy, Environment, Natural Resources, and Weather II, Session, Room 3. . . . . . . . . . . . . . . .50

Francisco Silva Pinto A framework for suitable prices: the role of alternative tariff structures in wa-ter supply projects

William Machado Emiliano Bus fleet management optimization using the augmented weightedTchebycheff method

Ricardo Lima Risk-averse stochastic programming vs robust optimization and sample average ap-proximation for a virtual power plantThu.3.4 Metaheuristics I, Session, Room 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51

Carlos Iglésias Solving large-scale railway line planning problems using a genetic algorithmXavier Andrade An heuristic approach to product line trimming decisionsAna Sofia Carvalho Ambulance dispatching and relocation problemAna Paula Teixeira Diabetes: Introduction to the analysis of the temperature of the foot

Thu.3.5 Discrete Optimization, Graphs, and Geometry II, Session, Room 5 . . . . . . . . . . . . . . . . . . . . . . . . 53

Telmo Pinto Models and advanced optimization algorithms for the integrated management of lo-gistics operations

Anabela Costa Heuristic algorithms for a capital budgeting problem with the option to deferPedro Rocha Reducing geometric complexity of irregular pieces for nesting problems.Marta Castilho Gomes New approach for optimization of construction and demolition waste man-

agement: Application to the Lisbon metropolitan areaThu.3.6 OR Teaching, Session, Room 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Susana Fernandes GLP-Tool: uma ferramenta gráfica, dinâmica e interativa para introdução àprogramação linear

João Miranda Estudo europeu em investigação operacional no ensino superior: resultados impor-tantes e tendências preliminares

José Soeiro Ferreira Investigação operacional e artes marciais

Friday, June 30, 9:30-10:50

Fri.1.1 Scheduling and Sequencing II, Session, Room 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Mariana Nogueira Scheduling of nonconforming devices: the case of a company in the automotivesector

Arminda Pata Aplicação de meta-heuristicas para minimização de lesões músculo esqueléticas rela-cionadas com o trabalho

Nicolau Santos Simultaneous management of sequence dependent setups and work in progress in aflexible job shop

Ana I. Pereira Optimal procedures to schedule home care visits in a health centerFri.1.2 Optimization and Multiple Parties, Session, Room 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

Aydin Teymourifar An analysis of the effects of pricing and government subsidies on patients’ hos-pital choices and social utility through simulation

Sara Martins MCVRP with product-specific time-windows assignmentZhor Chergui Reference points, which impact can have upon the reliability of the MCDM methods,

TOPSIS methods for Group decision makers & interval data : study and comparisonJoão P. Almeida Hotelling game on networks

IO 2017 13

PROGRAM OVERVIEW

Fri.1.3 Optimal Control and Systems Dynamics, Session, Room 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Carlos Balsa A direct approach to a control problem in passive tracer advection by viscous point

vortex flowOlena Kostylenko Banking risk as an epidemiological model: an optimal control approachRicardo Cunha Agile real-time decision makingHelena Sofia Rodrigues Epidemiological models and the diffusion of an advertising campaign

Fri.1.4 Metaheuristics II, Session, Room 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Parisa Sadeghi Balancing mixed-model assembly systems with a variable neighbourhood descent

methodNelson Chibeles-Martins Abordagem heurística na avaliação no sumódulo de risco de incêndio em

Solvência IICláudio Alves Variable neighborhood search for integrated planning and scheduling problems

Fri.1.5 Simulation and Stochastic Programming, Session, Room 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Luís Pereira The level of detail in simulation models: why does it matter?Ruben Dias A simulation and optimization framework for integrated planning and process designMário Amorim Lopes A stochastic agent-based model with optimization strategies for nursing

workforce planningRicardo Lucas The PackNStore case study

Fri.1.6 Augusto Queiróz Novais - Process Systems Engineering, Session, Room 6. . . . . . . . . . . . . . . . . . .63Miguel Vieira A decision-support framework for the planning optimisation of biopharmaceutical

industrial processesCatarina Marques A simulation-optimization framework to address the stochastic nature of process

design and planning decisions in the chemical-pharmaceutical industryPedro Castro Continuous-time formulations for multiproduct pipeline schedulingHelena Paulo Integrating stochastic programming with scenario reduction methods to support the

design of biomass to bioenergy supply chains under uncertainty

14 IO 2017

Plenary Sessions

ANA PAULA BARBOSA PÓVOA

Instituto Superior Tecnico, Universidade de Lisboa, apovoa@tecnico.ulisboa.pt

OPTIMISING SUSTAINABLE SUPPLY CHAIN

Wednesday, June 28, 15:00-16:00, AuditoriumChair: A. Ismael F. Vaz

Supply chain are complex systems involving a set of entities located around the world andwhose main objective is the supply of products to the final consumers at a minimum cost.The management of such systems is complex due to the large multiplicity of materials andinformation involved and the varied characteristics of the associated entities. In any or-ganization supply chain play a crucial role as they dictate the availability of the productsto the final consumer and consequently the financial success of the involved companies.Consequently, supply chain ought to be flexible, resilient and efficient while guaranteeingcostumers demands at minimum costs. Additionally, as supply chains are often associatedwith a high level of natural resources consumption and may be linked to high environmentand societal burdens a special care should be devoted to sustainability concerns when de-signing, planning and operating such systems. The emergence of the sustainable supplychain area is nowadays a reality in both academic and industry communities where severalchallenges are still to be resolved. To help answering some of these challenges the usage ofoperations research methods is as a path to follow as they allow the support of supply chaindecisions while accounting for sustainability concerns. In this talk we explore the use ofoperations research methods, more precisely optimisation, to address the design, planningand operation of supply chain towards the management of sustainable supply chain.

Ana Barbosa-Póvoa is a Full Professor of Operations and Logistics at the Department of Manage-ment and Engineering of Instituto Superior Técnico (IST), University of Lisbon, She holds a PhDfrom Imperial College of Science Tecnology and Medicine. Her research interests are on the areas ofOperations and Supply Chain Management where she has been exploring the optimization of globalsystems accounting for economic, environmental and social concerns. Ana has published widelyin these areas and supervised several Master and PhD students. She has been coordinating severalscientific projects some of them in close collaboration with industry.

IO 2017 15

PLENARY SESSIONS

RUTH KAUFMAN FORSThe OR Society, ruth.kaufman@btinternet.com

OR PRO BONO: THE UK EXPERIENCE

Thursday, June 29, 14:30-15:30, AuditoriumChair: José Fernando Oliveira

It is likely that for as long as O.R. has existed, O.R. people have given their time and skillsfreely, to friends, individuals, organisations or causes that they want to help. Several yearsago, the UK’s O.R. Society decided to formalise and expand this, by setting up a “Pro BonoO.R.” scheme supported and funded by the OR Society. Since its beginnings in 2010, thescheme has gradually grown: in 2016, non-profit organisations received an estimated 160person-days from 60 O.R. volunteers, and 38 new projects were initiated. In this talk RuthKaufman, who was one of the founder members of the scheme and is currently President ofthe OR Society, will talk about the UK experience, including: the history of the UK schemeand how it works, what sort of projects have been undertaken under the banner of Pro BonoO.R., the advantages and opportunities of such a scheme, and the risks, issues and chal-lenges that it raises.

Ruth Kaufman, a ‘Companion’ of the OR Society, became President of the Society in January 2016.Like many other members of the OR Society, she fell into O.R. by accident, in her case having takena maths BA in the School of Social Sciences at Sussex University. This led to a long career in publicsector O.R. and wider management at London Transport, London Electricity, Department of Healthand Export Credits Guarantee Department (ECGD). At ECGD, she joined the Executive Board, havingresponsibility for strategy and change management as well as leading an influential OR group. Inthe voluntary sector, Ruth chaired a small charity, Woman’s Trust, for five years and was a foundermember of the OR Society’s Pro Bono Scheme. Ruth is currently (amongst other things), Advisor tothe Finance Committee at the National Federation of Women’s Institutes, a visiting Senior Fellow atthe London School of Economics and a freelance consultant and advisor. She was awarded an OBEfor ’services to Operational Research’ in the 2016 New Year’s Honours list.

RICARDO VIANA VARGAS

Executive Director Brightline Initiative, ricardo@ricardo-vargas.com

LEADING PROJECTS IN TURBULENT TIMES

Friday, June 30, 11:20-12:20, AuditoriumChair: Lia Oliveira

Using real life examples, Ricardo will talk about the challenges faced in projects globally dueto the unstable environment and increasing volatility. He will share his leadership experi-ence in handling project in the most complex environments worldwide.

Over the past 20 years, he has been responsible for more than 80 major transformation projects inseveral countries within the oil and gas, energy, infrastructure, telecommunications, informationtechnology and finance industries; covering an investment portfolio of over 20 billion USD.

16 IO 2017

PLENARY SESSIONS

He is currently the Executive Director of the Brightline Initiative(TM). The movement is a coali-tion of leading global organizations from business, government and not for profit sectors, includingthe Boston Consulting Group (BCG), the Project Management Institute (PMI) and the Agile Alliance.

Brightline’s mission is to provide a knowledge and networking platform that delivers insightsand solutions to successfully bridge the gap between strategy development and strategy implemen-tation.

From 2012 to 2016, Ricardo Global Director of the Infrastructure and Project Management Groupwith the United Nations Office for Project Services (UNOPS) in Copenhagen, Denmark.

With a UNOPS portfolio of more than 1.2 billion dollars in projects, his work was primar-ily focused on improving the design, infrastructure and project management of humanitarian,peace-building and development projects all across the world, including but not limited to, Haiti,Afghanistan, Myanmar, Iraq and South Sudan.

Between 2007 and 2009, Ricardo served at the Board of Director and was the first Latin Americanto be elected Chairman of the Board of the Project Management Institute.

PMI is the largest project management organization in the world, with close to 2.9 million mem-bers and certified professionals.

Vargas is a recognized reviewer of the PMI’s A Guide to the Project Management Body of Knowl-edge (PMBOK(C) Guide), the most relevant reference manual in the world for project managementand also chaired its first official translation to Brazilian Portuguese.

Ricardo Vargas has written fifteen books on the subject of project management, published inPortuguese and English with translations in French, Spanish, Arabic, Chinese, and Danish whichhave sold over 300,000 copies globally.

He is also the host of one of the most relevant podcasts in the field, the 5 Minutes PM Podcast,published in English and Portuguese, with more than 4 millions views.

IO 2017 17

Sessions

Wed.0.0, Wednesday, June 28, 16:00-16:20, Au-ditorium

OR Pro BonoSession chair: Sandra Silva

1. "Rir é o Melhor Remédio?": Estudo do Im-pacto da Intervenção dos Doutores Palhaçosda Operação Nariz VermelhoC. Hiolanda Esteves (hiolan-daesteves@gmail.com) Universidade doMinho, C. Hiolanda Esteves, Susana Caires

Keywords: Doutores Palhaços; pediatria; human-

ização

Nos últimos anos, os grupos de Palhaços de Hos-pital têm-se expandido em vários países e a suaimportância tem vindo a ser reconhecida pelaSociedade. A observação da sua prática permitereconhecer os benefícios destas intervenções nacriança, reconhecimento esse que tem vindo aalargar-se à comunidade científica, no seio daqual estudos mais recentes apontam o riso e obrincar como elementos importantes no bem-estar e recuperação dos pacientes pediátricos. Oseu impacto é também perceptível ao nível dosseus familiares e profissionais de Saúde. Numatentativa de avaliar de modo mais sistemático assuas práticas de intervenção e os processos deformação dos seus artistas, mais recentemente,a Operação Nariz Vermelho desenhou, em co-laboração com o Instituto de Educação da Uni-versidade do Minho, o Projecto de investigação"Rir é o melhor remédio?". De entre os seus prin-cipais objetivos, destaca-se a intenção de avaliaro impacto físico, emocional, psicossocial e or-ganizacional da intervenção dos Doutores Pal-haços junto das crianças/adolescentes, famil-iares e profissionais de Saúde.

Wed.1.1, Wednesday, June 28, 16:50-18:30,Room 1

EstudIOSession chair: A. Ismael F. Vaz

1. Planning home health care services - arouting and scheduling problemAna Raquel Aguiar (a.raquel.aguiar@ist.utl.pt)Instituto Superior Técnico, Universidade deLisboa, Ana Raquel Aguiar, Tânia R. P. Ramos,M. Isabel Gomes

Extended abstract: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71

2. Impressão tridimensional bimaterial porprocesso aditivoDaniel Bandeira(danielf.a.bandeira@gmail.com) Departa-mento de Matemática, Universidade deCoimbra, Daniel Bandeira, Marta Pascoal

Extended abstract: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .73

3. Sistema de apoio à decisão para elaboraçãode mapas de exames no ensino superiorRaphael Cordeiro (ee12081@fe.up.pt) Facul-dade de Engenharia, Universidade do Porto,Raphael Cordeiro, Maria Antónia Carravilla,José Fernando Oliveira

Extended abstract: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75

4. Optimal control strategies for an advertise-ment viral diffusionJoão Gonçalves (jn-costagoncalves@gmail.com) AlgoritmiR&D Center, Department of Production andSystems, University of Minho, João Gonçalves,

IO 2017 19

SESSIONS

Helena Sofia Rodrigues, M. Teresa T. MonteiroExtended abstract: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78

5. Impacto da uniformização e centralizaçãode consumos em contexto hospitalarJuliana Marques (julianaacps@gmail.com) Es-cola de Economia e Gestão, Universidade doMinho, Juliana Marques, Nazaré Rego

Extended abstract: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .80

Wed.1.2, Wednesday, June 28, 16:50-18:30,Room 2

DEA and Performance AnalysisSession chair: Clara Bento Vaz

1. The evaluation of the social performanceof mining firmsRenata Oliveira (renata.oliveira@uepa.br) Fac-uldade de Engenharia da Universidade doPorto, Renata Oliveira, Andreia Zanella, AnaCamanho

Keywords: Composite indicators, Luenberger in-

dicator, social performance, mining firms

This study presents a composite indicator tosupport the evaluation of firms regarding socialachievements. In recent years, the social indica-tors of large corporations are increasingly beingused as a measure of their overall performanceregarding corporate social responsibility. Repu-tational issues associated with the firms’ impacton society, both in terms of local employmentand contribution to regional economic develop-ment are critical. This paper proposes a frame-work for the specification of indicators reflect-ing social burdens and benefits of mining firms,based on international standards and sectoralguidelines. A composite indicator is obtainedusing a Directional Distance Function model, al-lowing the identification of firms with the bestpractices in the sector regarding social aspects.The evolution of performance over time is as-sessed using the Luenberger indicator. An illus-trative application involving the assessment of24 large mining firm in the period 2011 to 2012is discussed. The managerial implications of the

results obtained are examined by the end of thiswork.

2. Malmquist and Hicks-Moorsteen produc-tivity indexes for clusters performance eval-uation

Rui Cunha Marques(rui.marques@tecnico.ulisboa.pt) CESUR,CERIS, Instituto Superior Técnico, Universityof Lisbon, Rui Cunha Marques, Diogo CunhaFerreira

Keywords: Cluster analysis; Malmquist index;

Hicks-Moorsteen index; Total Factor Productivity; un-

balancedeness; uncorrespondencedeness

Measuring the performance of clusters charac-terized by the unbalancedeness and units withno correspondence in other clusters ("uncor-respondencedeness") has not achieved the de-sired attention in the literature. Particularly,the operational research has been almost exclu-sively focused on performance evolution overtime, where clusters are generally balanced andthe units repeat themselves over these groups.Such analysis has been based on the Malmquistand the Hicks-Moorsteen indexes (MI and HMI),which are solely based on Shephard’s radial dis-tance functions and do not account for all ineffi-ciency sources. Making use of the so-called Geo-metric Distance Functions (GDF) and the GDF-based MI, we propose a generalization of theHMI, based on targets instead of distances to theefficient frontier, allowing the introduction of allinefficiency sources in the productivity model.We propose a Monte-Carlo-based framework toachieve the pseudo-corresponding units for gen-eral cluster performance analysis. This frame-work is a generalization of the conventional per-formance evolution over time. Wwe show thatthe HMI can be decomposed into economicallymeaningful indexes and can be rewritten as thegeometric mean of the input and the output-oriented MIs. Given these conclusions and ourproposed framework, the employment of theHMI to the general clusters analysis is straight-forward.

20 IO 2017

SESSIONS

3. Análise analítica da eficiência de rega us-ando modelos determinísticos e estocásticosRui Marques(rui.marques@tecnico.ulisboa.pt) Insti-tuto Superior Técnico, Helga Pereira, RuiMarques

Keywords: Eficiência do Uso da Água; Data Envel-

opment Analysis; Irrigação; Revisão da Literatura

Este trabalho tem como objetivo realizar uma re-visão da literatura sobre as metodologias apli-cadas para medir a eficiência do uso da águapara rega. A avaliação da eficiência das unidadesprodutivas é muito importante para fins es-tratégicos (comparação entre unidades produ-tivas), para planeamento (avaliação da utiliza-ção de diferentes combinações de fatores) e paraa tomada de decisões (para melhorar o desem-penho atual, analisando as diferenças entre aprodução atual e a potencial). Encontraram-se mais de trinta estudos que utilizaram méto-dos determinísticos e estocásticos para deter-minar a eficiência. Várias metodologias foramutilizadas, embora a técnica mais adotada foi oData Envelopment Analysis (DEA). Verificou-seque o DEA foi frequentemente utilizado junta-mente com análise de regressão para entendermelhor a fonte de ineficiência. Dos trinta e doisestudos analisados, 87,5% calcularam a eficiên-cia usando métodos determinísticos. Destes, ametodologia mais utilizada foi a DEA. Esses es-tudos levaram à conclusão de que, em geral, asexplorações são consideravelmente ineficientesem relação ao uso da água, e alguns fatores po-dem contribuir para sua melhoria, como por ex-emplo, escolha das culturas adequadas, ter umamaior área de exploração, ter um elevadoo nívelde educação, instalar rega localizada, ter acessoa melhores oportunidades de crédito.

4. Efficiency and Capital Structure in Por-tuguese SMEsAntónio Fernandes (antoniof@ipb.pt) AppliedManagement Research Unit (UNIAG); Poly-technic Institute of Bragança, António Fernan-des, Clara Bento Vaz, Ana Paula Monte

Keywords: Data Envelopment Analysis; Technical

Efficiency; Capital Structure; SME; Inland of Portugal

Acknowledgement: UNIAG, R&D unit funded

by the FCT - Portuguese Foundation for the Devel-

opment of Science and Technology, Ministry of Sci-

ence, Technology and Higher Education. Project no

UID/GES/4752/2016. NECE - Research Unit in Busi-

ness Sciences, Beira Interior University, Research unit

funded by the FCT - Portuguese Foundation for the De-

velopment of Science and Technology, Ministry of Sci-

ence, Technology and Higher Education. Project No

UID/GES/04630/201.

This paper aims to analyse the bi-directional re-lationship between technical efficiency, as mea-sure of companies’ performance, and capitalstructure, under the agency cost theory as wellas the pecking order and trade-off theory toexplain the capital structure decisions. Thetechnical efficiency was estimated by the DEAmethod and corrected by using a suitable boot-strap to obtain statistical inferences. To test theagency cost hypothesis, asymmetric informa-tion hypothesis, risk-efficiency hypothesis andfranchise value hypothesis (under pecking or-der and trade off theories framework) two mod-els were performed using some determinants ofcapital structure such as size, profitability, tangi-bility, liquidity as control and explanatory vari-ables through a truncated regression with boot-strapping. From an initial sample of 1024 smalland medium sized companies from the interiorof Portugal, for the period 2006-2009, it was se-lected a subsample of 210 SMEs from secondaryand tertiary sectors. The results suggest thatmedium sized companies have higher averagebias-corrected efficiency than small companies;short-term leverage is positively related to effi-ciency and the companies in the sample followpecking-order theory.

5. Determinants of nursing homes perfor-mance: The case of Portuguese Santas Casasda MisericórdiaAndré S. Veloso (asveloso@ipb.pt) PolytechnicInstitute of Bragança, André S. Veloso, ClaraBento Vaz, Jorge Alves

Keywords: Data Envelopment Analysis; Effi-

ciency; Nursing Homes; Third Sector

Acknowledgement: This work is financed by

the ERDF - European Regional Development Fund

through the Operational Programme for Com-

IO 2017 21

SESSIONS

petitiveness and Internationalisation - COMPETE

2020 Programme within project "POCI-01-0145-

FEDER-006961", and by National Funds through

the Portuguese funding agency, FCT - Fundação

para a Ciência e a Tecnologia as part of project

"UID/EEA/50014/2013".

This study aims to evaluate the economic effi-ciency of Nursing Homes owned by 96 SantasCasas da Misericórdia (SCM) and the determi-nants that influenced their efficiency in 2012 and2013. The SCM are the oldest non-profit entities,which belong to Third Sector in Portugal, pro-vide this social response and receive significantfinancial contributions annually from the state.The study is developed in two stages. In thefirst stage, the efficiency scores were calculatedthrough the non-parametric DEA technique. Inthe second stage, Tobit regression is used to ver-ify the effect of certain organizational variableson efficiency, namely the number of users andexistence of Nursing Home chains. The resultsof the DEA model show that the efficiency av-erage is 81.9%, and only 10 out of 96 NursingHomes are efficient. Tobit regression shows thatthe number of users has a positive effect on theefficiency of Nursing Homes, whereas the exis-tence of Nursing Home chains affects their effi-ciency negatively.

Wed.1.3, Wednesday, June 28, 16:50-18:30,Room 3

Production and Inventory Manage-mentSession chair: Paula Alves

1. A spreadsheet based model that estimatesthe impact of reduced distribution time oninventory investment savingsSerhat Saylam (serhat.saylam@metu.edu.tr)Middle East Technical University, Serhat Say-lam

Keywords: Fill rate; Lead time; Inventory cost;

Spreadsheet modeling; VBA, ABC Classification

This study proposes a spreadsheet model thatestimates the impact of reduced lead time on in-ventory investment savings. The study providesusers with a means of automatically calculat-

ing inventory control parameters such as safetystocks and reorder points, and estimating thesavings caused by lead time mean or variabil-ity reduction. Then a trade-off analysis can bedone to determine whether reducing lead timewould override the lead time crashing cost. Inthe model presented here, the model finds theoptimal safety factor of an item based on a fillrate goal using Excel Solver. Then, Excel’s VBAautomates the process of finding safety factorsfor other items before and after lead time reduc-tion. To see the functionality of the model andinterpret the results, it is applied to different sup-ply support activities of USTRANSCOM.

2. Solving the backroom design problem

Maria Pires (maria.pires@fe.up.pt) FEUP andINESC TEC, Maria Pires, Pedro Amorim

Keywords: Backroom design, Grocery retail,

Strategic Planning

The grocery retail environment is more dynamicthan ever and competition keeps intensifying.This requires retailers to adapt and develop in-novative approaches to face current challenges.However, fresh thinking concerning backroomsis rare, in both academia and practice. In thispresentation, we describe a methodology to de-sign grocery backrooms. The proposed method-ology consists in three stages: demand forecast-ing, backroom sizing and designing the back-room layout. The forecasting model is basedon clustering techniques and multinomial lo-gistic regressions. Furthermore, two modelswere developed for the backroom sizing prob-lem. The first formulation is a bottom-up ap-proach that aims to reduce the backroom life-cycle-costs by determining the optimal storagespace and height. The second is a top-down ap-proach based on DEA that determines the effi-cient level of floor space for each department.Lastly, a mathematical model to solve the back-room layout problem was developed. This prob-lem consists in determining the best location forthe departments in the backroom consideringtheir sizes, adjacencies, physical restrictions andlocation of the corresponding departments inthe sales area. The proposed methodology wastested on more than forty convenience stores,

22 IO 2017

SESSIONS

using data from a European retailer. The appli-cation of this methodology demonstrates a sub-stantial potential for space savings.

3. Review policies for online retail inven-tory management considering customer or-der time windowsJorge Mendes (em12099@fe.up.pt) INESC TECand FEUP, Jorge Mendes, Pedro Amorim,Gonçalo Figueira

Keywords: Online retail inventory management

The growth of e-commerce retail brings newchallenges and opportunities for retailers, par-ticularly in what concerns their inventory poli-cies. The delay between the time an order isreceived and the time at which the customerwants it delivered provides an order window thatallows additional flexibility for the retailer. Inthis work, we present an (s,Q) inventory policythat explicitly accounts for the ordering windowwhen minimizing ordering, holding and stock-out costs. Considering the flexibility providedby the ordering window we find expressions forthe optimal parameters of the policy. We con-sider that the order arrival as well as the order-ing window are stochastic and we explore differ-ent probability distributions. The optimal pol-icy is obtained considering both single and mul-tiple on-order scenarios and it is validated usingdiscrete-event simulation. The results obtainedwith this policy may be very useful as they wouldallow significant savings for retailers for a givenservice level. Nevertheless, there are still severalproblem extensions to be explored in this onlineretail setting.

4. A prescriptive analytics method appliedto a maintenance and manpower schedulingproblemJoão Correia Pinto (jcpinto@inesctec.pt) IN-ESC TEC, João Correia Pinto, Rebelo de SáCláudio, Carlos Soares, Hugo Ferreira

Keywords: maintenance, prescriptive analytics,

scheduling, data mining, machine learning

In this paper we propose a framework whichcombines the three phases of analytics (descrip-tive, predictive and prescriptive) for mainte-nance problems. In simple terms, this tool is

able to allocate the workforce to particular main-tenance jobs, in a timely fashion, to avoid unex-pected failures and minimize the total costs. Toachieve that, a machine learning model is pre-viously trained and tested to predict the failuretimes of a set of machines. Based on these pre-dictions, a scheduling model allocates the avail-able workforce in an cost-optimal manner. Ourapproach takes into account the differentiatedcosts of maintenance that machines require be-fore and after the failures occur. As a proof ofconcept, a case study is analyzed where we showthe efficiency and applicability of the solution.

5. Desenvolvimento de uma metodologiapara redução do volume de obsoletos numaempresa do sector automóvel

Paula Alves (paulamarujo.alves@gmail.com)Escola Superior Ciências Empresariais - IPVC,Paula Alves, Lia Oliveira, Jorge Esparteiro Gar-cia

Keywords: Mapeamento, Obsoletos, optimização

Alguns produtos, com o passar do tempo deixamde ser necessários ou caem em desuso, resul-tado de uma possível melhoria ou substituição,fazendo com que os mesmos se tornem obso-letos, esta é uma realidade comum a qualquerempresa do ramo automóvel. Por esse mo-tivo, as empresas contam com um leque de ma-teriais em stock sem uso, que são em grandeparte sinónimo de perdas económicas. Efecti-vamente, os materiais obsoletos perdem o seupropósito original e dificilmente poderão serconvertidos em receita e implicam consequên-cias económicas negativas, seja no que diz re-speito à ocupação de espaço no armazém ou aeventuais custos operacionais. Foi neste sen-tido que surgiu o projeto de desenvolver umametodologia para redução do volume de obso-letos na empresa alvo de estudo. Numa primeirafase, foi desenvolvido um processo que permi-tiu a identificação e consequente tratamento dosobsoletos, fazendo todo o mapeamento do pro-cesso. Com o intuito de melhorar a eficiên-cia do procedimento foi feito um levantamentode soluções informáticas de forma a automati-zar e a melhorar a eficiência do processo desen-volvido. Os resultados obtidos, provaram a vali-

IO 2017 23

SESSIONS

dade da metodologia reduziram em mais de 10%o valor de obsoletos identificados.

Wed.1.4, Wednesday, June 28, 16:50-18:30,Room 4

Data Analysis and Artificial In-teligenceSession chair: Helena Alvelos

1. The two-dimensional strip packing prob-lem: what matters?

Alvaro Neuenfeldt Júnior (al-varo.l.junior@inesctec.pt) INESC-TEC,Faculty of Engineering, University of Porto,Alvaro Neuenfeldt Júnior, Elsa Silva, A. MiguelGomes, José Fernando Oliveira

Keywords: Strip packing problem; Cutting and

packing problem; Principal component analysis;

Knowledge discovery

In this paper, an exploratory approach is con-ducted to deeper understand the most signifi-cant characteristics of the two-dimensional strippacking problem. A number of variables are de-fined to represent the main problem character-istics, segregated into four groups, establishedthrough qualitative knowledge about the con-text of the problem and considering referenceparameters used to develop each variable. Atthe same time, a linear correlation is used as aquantitative measure to validate the assignmentof variables to groups. In a second moment,principal component analysis (PCA) is used toreduce the dimensions of each group, revealingthe relationship between each variable and thecomponents generated by the PCA. Our analy-sis indicates that the problem can be reduced toten characteristics, retaining most part of the to-tal variance explained by the sampling. Thesecomponents can be used to fit parametric ornon-parametric regression models to estimatethe strip height necessary to position all itemsinside the strip.

2. The reverse logistics of unsold medicationsin pharmacies in Campania, Italy

Isabel Cristina Lopes (cristi-nalopes@iscap.ipp.pt) ISCAP - P.Porto,Rosekelly Araújo Costa, Teresa Pereira, Isabel

Cristina Lopes

Keywords: Reverse Logistics, Expired date medi-

cations, Pharmacies, Region of Campania-Italy, Factor

Analysis

This paper is a study in Reverse Logistics (RL)that aims to analyse the reverse flow of medi-cations with expired dates, in the pharmacies ofthe Campania region in Italy. The main objectiveis to analyse the final destination of medicationsthat are not sold and are collected in pharmacies.The analysis of how the company responsible forthe collection of the medicines works was madeusing semistructured interviews, and a subse-quent factor analysis of the collected data. Thepharmacies of the main cities of this region wereinvestigated, in order to understand their impor-tance in this process, as well as to understandtheir main difficulties and challenges. A statis-tical analysis of the data allowed us to verify howpharmacies are accustomed to the current legis-lation and are aware of the importance of theirrole in the RL of the medications that are notsold with the expiration date. It was observedthat pharmacies are very satisfied with the com-pany responsible for the collection and referralof medicines and their materials to an adequatefinal destination. Both of them work in tune, re-spond well to current legislation and respect theenvironment.

3. Optimization of short-term storage of con-tainers in empty container terminals

José Oliveira(jose.s.oliveira@tecnico.ulisboa.pt) CERIS,CESUR, Instituto Superior Técnico, Univer-sidade de Lisboa, José Oliveira, Rui Oliveira,Marta Castilho Gomes, Vasco Reis, RosárioMacário

Keywords: Space allocation, empty container ter-

minals, heuristics, uncertainty modelling

Small scale empty container terminals are sub-ject to severe space and operational constraints.The need to stack containers belonging to dif-ferent clients and bookings in the same stackcomes at the expense of higher operational costsand lower service level. The decision on where tostore a newly arrived container has thus a major

24 IO 2017

SESSIONS

impact. In Lisbon Multipurpose Terminal (LMT)of Sociedade Portuguesa de Contentores (SPC),due to legal and operational environment con-ditions, the need for a specific container to beavailable at an unknown time in the future is themain source of uncertainty. A process for op-timizing the space allocation for a container isproposed for this case study, which relies on de-termining several levels of uncertainty. These aredefined by breaking down the available informa-tion on the parameters of the statistical distribu-tion of containers that enter LMT. Accordingly,the definition of the yard allocable space is trick-led down until an appropriate position is found,i.e. a position that minimizes the number ofmachine-hours spent on moving other contain-ers. Results are presented regarding the classifi-cation of uncertainty in containers’ dwell time.Moreover, the implementation of a collection ofheuristics towards the optimization of space al-location is proposed.

4. Automatic learning of planning rules foroperations management problems

Gonçalo Figueira (goncalo.figueira@fe.up.pt)INESC TEC and FEUP, Gonçalo Figueira,Fanny Hein, Christian Almeder, Rui Lopes, Pe-dro Amorim, Bernardo Almada-Lobo

Keywords: Production planning, inventory man-

agement, planning rules, automatic learning, genetic

programming

When compared to mathematical programming,planning rules have important advantages: im-proved usability (do not require commercialsolver licenses); more easily understood by prac-titioners (unlike MP models/solvers which areblack-boxes); typically more consistent in gen-erating similar solutions (less disruptive); and,most importantly, extremely faster (enabling theoptimization of large, complex and stochasticproblems in real-time). Rules can be devisedin different ways, and that will highly dependon the problem. When mathematical deriva-tion and problem analysis are not practical, au-tomatic learning may be the answer. We presentsome examples of rules for operations man-agement problems, which can be learned orevolved automatically. Automatic learning is dis-

cussed at different levels (parameters, expres-sions and structure) and in the perspective ofvarious methods (evolutionary algorithms, arti-ficial neural networks, decision trees and geneticprogramming). These methods and perspectivesare instantiated in different management prob-lems, particularly production planning, inven-tory management and vehicle routing. Geneticprogramming appears to be one of the most gen-eral, powerful and transparent approaches toobtain rules, although it may be too ambitiousor even overkill in some cases.

5. Melhoria do processo de corte a laser numaempresa metalomecânica

Helena Alvelos (helena.alvelos@ua.pt) Univer-sity of Aveiro, Helena Alvelos, Hugo Maurício

Keywords: Qualidade; Corte a laser; Chapa

metálica; Análise de variância.

Neste trabalho apresenta-se um estudo cujo ob-jetivo é o de analisar os factores que influenciamo corte a laser de placas de chapa metálica, nosentido de propor ações de melhoria que per-mitam diminuir os desvios entre das cotas daspeças produzidas relativamente às suas especi-ficações. Para tal, foram recolhidas amostras daprodução após o corte a laser, tendo sido uti-lizadas metodologias da qualidade para anal-isar os dados recolhidos e identificar as causasdo problema, nomeadamente o diagrama deIshikawa e os histogramas. Foram identifica-dos fatores que potencialmente afetam o desvioreferido e foram testados os seus efeitos, uti-lizando a técnica de Análise de Variância. Osfatores considerados foram a posição de ar-mazenamento da chapa (na vertical ou na hori-zontal), a matéria prima (ferro ou aço) e a espes-sura da peça (menor do que 2mm, entre 2mm e5mm e maior ou igual do que 5mm). Foi, ainda,feita uma análise de correlação entre a variáveisdesvio relativamente à especificação e a especifi-cação propriamente dita. A partir dos resultadosobtidos foram retiradas conclusões e apontadasalgumas sugestões de melhoria.

IO 2017 25

SESSIONS

Wed.1.6, Wednesday, June 28, 16:50-18:30,Room 6

Logistics and Supply ChainSession chair: Ana Paula Barbosa Póvoa

1. Green supply chain design and planning:the importance of decision integration in op-timization modelsBruna Mota (bruna.mota@tecnico.ulisboa.pt)Instituto Superior Tecnico, Bruna Mota, AnaCarvalho, Maria Isabel Gomes, Ana Paula Bar-bosa Póvoa

Keywords: supply chain, optimization, sustain-

ability, decision integration, Life Cycle Assessment

Sustainability is more than ever a central con-cern when deciding a company’s strategy. Stake-holders are continuously pressuring industriesto further reduce their environmental impact. Inorder to achieve that, it is imperative to look forsolutions across all of the company’s operationsadopting a supply chain view. However, sup-ply chains’ complexity makes this a challengingtask. Analyzing the sustainability of such com-plex systems requires the use of capable tools asare optimization models. Several of these mod-els can be found in the literature, mostly focus-ing on specific issues or specific decisions in thesupply chain. Therefore, a research gap is foundin models capable of handling a wider variety ofdecisions. With this work a mixed integer lin-ear programming model is used to demonstratethe impact of decision integration on design andplanning decisions, and on the environmentalperformance of a supply chain. A case-studybased on a Portuguese pulp and paper companyis analysed.

2. Leveraging logistics flows to improve thesludge management process of wastewatertreatment plantsPedro Amorim (pamorim@fe.up.pt) FEUP,Alda Henriques, Milton Fontes, Ana Camanho,Jaime Gabriel Silva, Pedro Amorim

Keywords: Wastewater treatment plants; sludge

management; logistic network optimization; biogas

production

This study considers a system-wide perspec-

tive for sludge flows management in wastewatertreatment plants (WWTP). Sludge has great en-ergetic potential, being the anaerobic digestion(AD) widely recognized as a very suitable tech-nology for its recovery. That’s why bigger WWTPdesign includes AD facilities. The input to AD isthe output that must be disposed at the end ofthe wastewater treatment process. Therefore, in-tegrating both processes can lead the enhance-ment of biogas production and facilitate sludgedisposal. This synergy is leveraged by the ex-istence of unutilized capacity in some digestersand the possibility of reducing the sludge ageduring the aerobic treatment in the plants with-out AD facilities, so that the sludge retains itsbiogas potential. In this context, we develop anoptimized network to connect the plants with-out AD facilities to the ones with anaerobic di-gesters. This follows other studies of biomasssupply chain design, specifically considering theminimization of biogas loss during the logisticsoperation of sludge and the maximization of theoverall revenues. The resulting framework offersnew insights for an optimized sludge manage-ment process of WWTP, contributing to energyself-sufficiency in the system. The approach de-veloped is applied to a Portuguese company withseveral plants.

3. Supply chain modelling and supplier selec-tion in the aerospace industry

Nuno Falcão e Cunha (nuno.cunha@fe.up.pt)FEUP, Nuno Falcão e Cunha, BernardoAlmada-Lobo, Pedro Amorim

Keywords: Supply Chain Management, Supplier

Selection, Aerospace

The aerospace sector poses many unique chal-lenges as companies competing in this indus-try become increasingly globalized. The devel-opment of a single aircraft model can take upto 15 years, including R&D, design, material se-lection and certification, which must all be ac-companied with the assembly of a suitable sup-ply chain capable of assembling and deliveringthe final product. To study operational trade-offsassociated with decisions made in supply chaindesign, we propose an analytical model describ-ing the interaction between relevant stakehold-

26 IO 2017

SESSIONS

ers within the supply chain, including multi-ple supplier tiers, manufacturing and assemblysites, as well as end-customers. We will studythe impact of varying production volumes, dif-ferent processes and manufacturing practices,as well as the modularity of the product and cor-responding supply chain. This model will yieldpropositions and practices for efficient supplychain development. Furthermore, we will de-velop a decision support system capable of iso-lating the critical metrics for supplier selection.Using techniques associated with complex deci-sion making, such as AHP, and building on con-tacts with key informants from the industry, aswell as the propositions and policies from theglobal supply chain model described in the pre-vious section, we aim to enhance supplier selec-tion in the aerospace sector.

4. Pharmaceutical Supply Chain: a com-prehensive optimal planning for medicineshortage

Ana Cristina Amaro (ISCAC| Coimbra BusinessSchool) aamaro@iscac.pt, Ana Cristina Amaro,Ana Paula Barbosa-Póvoa

Keywords: Optimal Planning, Supply Chain,

Pharmaceutical, medicine shortage

Large business challenges are being faced bycompanies as a result of some unpredictablevariability observed in demands, prices, cus-tomer expectations and product life cycles.These explains, in somehow, the behaviorof trade relations, industrial investments andglobal business positions. Risk and uncertaintyhad a particular impact in the pharmaceuticalindustry due to the strengthening complianceand regulatory requirements that have to bemanaged. Pharma companies entered the Glob-alization "Era" with a recognized investmenton partnership collaborations and alliances tomaintain a healthy sustainable position. Be-sides that, the growing of pharmaceutical ex-penditure emerged in a scenario of short-termrigid cost-containment measures. Accordingly,generic pharma and biosimilar medicines hada noticeable industrial impact. In this contri-bution the optimal planning of industrial phar-maceutical Supply Chains (SC) is addressed. A

comprehensive approach based on the Chain-STN representation is analyzed in depth. A Mas-ter representation supported on Events Replica-tion is proposed to optimally model the pharma-ceutical SC planning. A sustainable economicobjective is coupled with customer complianceand shortage criteria, while accounting for oper-ational planning conditions. The proposal andthe associated planning methodology are illus-trated through the solution of an industrial case-study and the achieved results are appreciatedand discussed.

5. Design and planning of sustainable supplychains - the case study of a tissue paper busi-ness

Ana Paula Barbosa Póvoa(apovoa@tecnico.ulisboa.pt) Instituto Su-perior Tecnico, Bruno van Zeller, Bruna Mota,Ana Paula Barbosa Póvoa

Keywords: supply chain, design and planning,

sustainability, supply chain modelling, optimization,

postponement

While planning to expand its tissue paper busi-ness a Portuguese company aims to explore op-tions regarding the design and planning of itssupply chain accounting not only for economicobjectives but also environmental and socialconcerns (the three pillars of sustainability). Amulti-objective mixed integer linear program-ming model (MOMILP) is developed and appliedto the company supply chain focusing on thesupply of tissue paper to the United Kingdommarket. Decisions to be taken include the net-work structure, entity location and capacity def-inition, transportation network definition, pro-duction and storage levels, and material flowplanning. Additionally it is important to de-cide whether or not to postpone the conver-sion to the final product to a location closer tothe clients. The augmented epsilon-constraintmethod is applied to study the trade-off betweenthe three pillars of sustainability and importantmanagerial insights are derived from this study.

IO 2017 27

SESSIONS

Thu.1.1, Thursday, June 29, 9:30-10:50, Room1

Other Models, Methods, and Algo-rithmsSession chair: Maria João Alves

1. Understanding complexity in a practicalcombinatorial problem using MathematicalProgramming and Constraint ProgrammingMaria Antónia Carravilla (mac@fe.up.pt) IN-ESC TEC and Faculty of Engineering, Univer-sity of Porto, Beatriz Brito Oliveira, Maria An-tónia Carravilla

Keywords: Job-shop scheduling problem; math-

ematical programming; constraint programming;

global constraints; complexity

Acknowledgement: The first author was sup-

ported by grant SFRH/BD/103362/2014 from FCT -

Fundação para a Ciência e Tecnologia (Portuguese

Foundation for Science and Technology). This

work was also partially financed by the ERDF -

European Regional Development Fund through the

Operational Programme for Competitiveness and

Internationalisation - COMPETE 2020 Programme

within project "POCI-01-0145-FEDER-006961", and

by National Funds through the FCT - Fundação

para a Ciência e Tecnologia (Portuguese Founda-

tion for Science and Technology) as part of project

UID/EEA/50014/2013.

Optimization problems that are motivated byreal-world settings are often complex to solve.Bridging the gap between theory and practice inthis field starts by understanding the causes ofcomplexity of each problem and measuring itsimpact in order to make better decisions on ap-proaches and methods. The Job-Shop Schedul-ing Problem (JSSP) is a well-known complexcombinatorial problem with several industrialapplications. This problem is used to analysewhat makes some instances difficult to solve fora commonly used solution approach - Math-ematical Integer Programming (MIP) - and tocompare the power of an alternative approach:Constraint Programming (CP). The causes ofcomplexity are analysed and compared for bothapproaches and a measure of MIP complexity

is proposed, based on the concept of load permachine. Also, the impact of problem-specificglobal constraints in CP modelling is analysed,making proof of the industrial practical inter-est of commercially available CP models for theJSSP.

2. A formulation for the multi-row layoutproblemManuel Vieira (mvcv@fct.unl.pt) UniversidadeNova de Lisboa, Manuel Vieira, Miguel Anjos

Keywords: facility layout, MILP

The multi-row facility layout problem (MRFLP)is the most general version of row layout prob-lems. An instance of the MRFLP has a givennumber of rows to which the machines can beassigned, the machines all have the same height(equal to the row height), the distances betweenadjacent rows are equal, and machines can ingeneral be assigned to any row. We present herea mixed integer linear formulation which is con-tinuous in both dimensions x and y, where x rep-resents the position within rows and y returnsthe row assigned to each machine. Despite, y be-ing a continuous variable, optimal solutions areattained for y integer. We present our numericalresults compared against other published for-mulations.

3. Ilustração de diferentes tipos de soluçõesem problemas de programação em doisníveis semivetorialMaria João Alves (mjalves@fe.uc.pt) CeBERe Faculdade de Economia, Universidade deCoimbra / INESC Coimbra, Maria João Alves,Carlos Henggeler Antunes

Keywords: programação em dois níveis semiveto-

rial; solução otimista; solução pessimista; solução de-

cecionante, solução recompensadora

A programação em dois níveis (bilevel) permitemodelar problemas de otimização hierárquicacom dois níveis de decisão. O líder (decisor denível superior) e o seguidor (decisor de nível in-ferior) controlam diferentes variáveis e procu-ram otimizar os seus objetivos de uma formanão cooperativa e sequencial, estando sujeitosa restrições que envolvem os dois conjuntos devariáveis. Um problema em dois níveis semive-

28 IO 2017

SESSIONS

torial tem uma única função objetivo no nívelsuperior e múltiplas funções objetivo no nívelinferior. A existência de múltiplas funções ob-jetivo no nível inferior dá origem a um con-junto de soluções eficientes para cada decisãodo líder, o que causa dificuldades adicionais aolíder em antecipar a reação do seguidor. Con-sequentemente, devem ser calculados diferentestipos de soluções de modo a apoiar o líderna sua decisão, fornecendo-lhe informação so-bre possíveis resultados e intervalos de valorespara a sua função objetivo para diferentes de-cisões do seguidor. Nesta apresentação serãointroduzidos dois novos tipos de soluções (de-cecionante e recompensadora) que, juntamentecom a solução otimista e a solução pessimista,serão ilustrados usando exemplos gráficos. Estesquatro tipos de soluções representam possíveisresultados "extremos" do processo de decisão,captando distintas posições do líder e reações doseguidor.

Thu.1.2, Thursday, June 29, 9:30-10:50, Room2

Multicriteria ISession chair: Luis C. Dias

1. Abordagem multicritério ao problema deexpansão de um sistema produtor elétricoManuel Matos (mmatos@inesctec.pt) INESCTEC & FEUP, Leonel Carvalho, Pedro Barbeiro,Manuel Matos

Keywords: Multicritério, Fiabilidade, Sistemas

elétricos

Acknowledgement: Este trabalho é financiado

por Fundos FEDER através do Programa Operacional

Competitividade e Internacionalização - COMPETE

2020 no âmbito do projeto POCI-01-0145-FEDER-

006961 e por Fundos Nacionais através da FCT - Fun-

dação para a Ciência e a Tecnologia através do projeto

UID/EEA/50014/2013

A expansão dos sistemas de produção de eletri-cidade é genericamente deixada ao cuidado demecanismos de mercado, com monitorização dasegurança de abastecimento por uma entidadepública que, em face de riscos excessivos, poderecomendar investimentos. No entanto, há sis-

temas onde se realizam exercícios de planea-mento, seja porque a atividade de produção nãoestá entregue ao mercado, seja porque se trata desistemas isolados. Nesta apresentação descreve-se um estudo incluído no processo de planea-mento de um desses sistemas, em que se iden-tifica um conjunto de soluções de Pareto tendoem conta critérios económicos (custos de in-vestimento e exploração), segurança de abastec-imento (indicadores de fiabilidade) e partici-pação de energia renovável na produção to-tal. As soluções (nº de novas máquinas a in-stalar em cada ano do horizonte de estudo) sãogeradas recorrendo à meta-heurística NSGA-II,complementada com uma metodologia de cál-culo rápido de índices de fiabilidade. Os resulta-dos permitem aos planeadores uma visão globaldas alternativas, sendo possível introduzir re-strições e regras de decisão a posteriori para lim-itação de conjunto de soluções a analisar. A faseseguinte do processo, que já não é objeto destaapresentação, consiste na análise das alternati-vas selecionadas em relação a outros aspetos, in-cluindo o impacto da rede.

2. Evaluating suppliers in the olive oil sectorusing AHP

Dalila B. M. Martins Fontes (fontes@fep.up.pt)Faculdade de Economia da Universidade doPorto, and LIAAD/INESC TEC, Dalila B. M.Martins Fontes, Teresa Pereira, Elisabete Dias

Keywords: Multi-criteria decision making, AHP,

olive oil sector

This work proposes a multi-criteria decisionmaking approach to help assessing and select-ing suppliers in the olive oil process. Olive oil isa protected agricultural product, by region andorigin certificate: protected Designation of Ori-gin, Protected Geographical Indication, and Tra-ditional Speciality Guaranteed. Therefore to se-lect a supplier, it is of utter importance to inspectand test (taste, colour, smell, density, amongothers) the olive oil in addition to the supplyingcompany. The identification of possible suppli-ers was done in two stages: firstly, the region oforigin from which to choose possible supplierswas identified and then potential suppliers wereevaluated on a set of characteristics for which

IO 2017 29

SESSIONS

minimum threshold values were set. From thisstudy, which is not part of the research reportedhere, we were able to identify the suppliers ofinterest. Due to the several characteristics andcharacteristic dimensions used to choose a sup-plier we resort to the Analytic Hierarchy Pro-cess to rank them, this way allowing for a betterchoice. The rank obtained is robust as the topranked supplier remains the same for any rea-sonable change in the criteria weighs and in theevaluation of the suppliers on each criterion.

3. The quality of service: An overall perfor-mance assessment for water utilities

Ana Sara Costa (anasara-costa@tecnico.ulisboa.pt) CEG-IST, InstitutoSuperior Técnico, Universidade de Lisboa,Francisco Silva Pinto, Ana Sara Costa, José RuiFigueira, Rui Cunha Marques

Keywords: Multicriteria; Decision support sys-

tems; Policy analysis; Quality of service; Water supply

services; ELECTRE TRI-nC

Over the last decades, the delivery of water sup-ply services has changed among regions and ac-tors, in order to meet the changing needs. Thoseservices can be balanced based on the relationbetween the obligations they must fulfill, thequality offered, and the associated price/valuefor money. Furthermore, due to the water ser-vices inherent characteristics, they are proneto poor quality of service (in general terms).Thus, global performance assessments are ofparamount relevance under those constraints.In Portugal, the water regulator has developed asystem of performance indicators (partial mea-sures of performance) and, therefore, it is notpossible to achieve the desired holistic perfor-mance evaluation (global measure). To enablesuch assessment, we propose an applicationbased on the ELECTRE TRI-nC method to definequality of service categories and aggregate per-formance indicators. To obtain a coherent fam-ily of criteria (i.e., exhaustive, cohesive, and non-redundant) we apply an iso-preference logic.The results obtained are presented through ageographical information system allowing for aclearer visualization of the overall performanceof water utilities. The method proposed can be

considered a suitable decision support systemand useful regulatory tool, able to provide policyrelevant outputs.

4. Uma abordagem de avaliação multicritériopara classificação e ordenação de alternati-vas conjugando Delphi e ELECTRE TRI

Luis C. Dias (lmcdias@fe.uc.pt) INESC Coim-bra, CeBER e Faculdade de Economia, Uni-versidade de Coimbra, Luis C. Dias, CarlosHenggeler Antunes, Guilherme Dantas, Ni-valde de Castro, Lucca Zamboni

Keywords: MCDA, ELECTRE, Delphi, Decisão em

grupo, Avaliação de políticas

Esta comunicação apresenta uma metodologiade avaliação multicritério e uma aplicação àavaliação de políticas de incentivo à inovaçãono setor elétrico, num contexto de evoluçãopara redes inteligentes (smart grids) no Brasil.Construiu-se uma hierarquia de objetivos no de-curso de uma abordagem estruturada para eli-ciar e organizar informação recolhida na liter-atura e em reuniões com especialistas do setor.Um conjunto de políticas de incentivo foi avali-ado por participantes (28) num processo Delphiquanto ao seu contributo para os vários obje-tivos, utilizando uma escala qualitativa. Os mes-mos participantes também se pronunciaram ac-erca da importância de cada objetivo, resultandoem informação qualitativa que foi traduzida soba forma de restrições aos pesos dos critérios deavaliação, sem necessidade de estabelecer re-lações de natureza cardinal entre os níveis qual-itativos indicados. As políticas foram objetode uma classificação, seguindo-se uma orde-nação das políticas em cada categoria. Para tal,considerou-se uma exploração estocástica e ex-ata da variação dos resultados da classificaçãoem função dos valores dos parâmetros quecumpriam as restrições estabelecidas. Foramobtidos resultados considerando o conjunto detodos os intervenientes e também para subcon-juntos que representavam diferentes perspeti-vas.

30 IO 2017

SESSIONS

Thu.1.3, Thursday, June 29, 9:30-10:50, Room3

MultiobjectiveSession chair: Sandra Silva

1. An efficient heuristic method for Paretooriented optimization of timecost tradeoffproblem for construction projects

Saman Aminbakhsh(saman.aminbakhsh@metu.edu.tr) MiddleEast Technical University, Saman Amin-bakhsh, Rifat Sonmez, Tankut Atan

Keywords: construction projects, time-cost trade-

off problem, optimization, project management,

scheduling

In construction projects, although it is desirableto accelerate the project schedule, it imposes ad-ditional direct cost while decreasing the dailyindirect cost. This trade-off between time andcost is known as the multi-objective time-costtrade-off problem (TCTP). The ultimate resolu-tion of TCTP is to find non-dominated set of so-lutions, named as the Pareto front, for the feasi-ble set of durations. Despite the plethora of re-search that has documented the significance ofthis problem, few have concentrated on design-ing methods for solution of large-scale projects.In this study, a novel Cost-Slope Heuristic is de-veloped which consists of unique techniques forexerting Pareto front optimization of large-scaleTCTPs. The proposed method initiating withthe all-crashed schedule, consists of two majorphases for identification of the uncrashable ac-tivities. While giving the uncrashing priority tothe activities with greater cost-slopes, the firstphase determines the activities which are exclu-sively on an excessively shortened path and thesecond phase determines the ones with effectivecost-slopes greater than or equal to daily indirectcost. The results of the computational experi-ments reveal effectiveness of this method sinceit is capable of providing high number of Paretofront solutions with minor deviations from theglobal optima.

2. Use of analytic hierarchy process (AHP) tosupport the decision-making about destina-

tion of a batch of defective products with al-ternatives of rework and discard

João Soares (id6293@alunos.uminho.pt) Cen-tre ALGORITMI, University of Minho, JoãoSoares, Anabela Tereso, Sérgio Sousa

Keywords: Analytic Hierarchy Process (AHP),

Cost, Decision-making, Quality

This study discusses the application of AHP tosupport the decision-making about destinationof a batch of defective products. The alterna-tives of destination are rework or discard. Sixcriteria of analysis and comparison were used.The mathematical development of the modelwas performed in Excel, which allowed severalinteractions and simulations, giving greater re-liability to its application. The study was de-veloped in a Brazilian plant of a Japanese autoparts industry which supplies a world-renownedJapanese motorcycle manufacturer. From a flowof analysis of quality problems, the AHP methodwas adapted and applied in this case study, us-ing evaluation questions to establish the criteriafor comparison. The problem analysis promotesanswers and determination of criteria weightsaccording to the influences of the answers on thecost and the quality of the product in each alter-native. The AHP method assisted the systemati-zation of the decision process, allowing the de-veloped system to be used in other quality prob-lems involving the destination of defective prod-ucts. The contribution of this work is the adap-tation of the AHP method to the application ofproblems of this type, using questions and an-swers. The applied method assisted in the deci-sion to discard the parts of the study.

3. Multiobjective irrigation model. TheAlqueva river basin.

Helga Pereira (hel-gapereira@tecnico.ulisboa.pt) Institutosuperior Técnico, Helga Pereira, Rui Marques,José Rui Figueira

Keywords: Multiobjective model; Irrigation;

Alqueva river basin.

This work aims to provide a multi-objective opti-mal water resources allocation model. The pro-posed model integrated integer programming

IO 2017 31

SESSIONS

and linear programming, using the Chebyshevmetric. The developed model was applied to ir-rigation water resources optimal allocation sys-tem Alqueva river basin, Alentejo. This study hasthree optimal objective functions: economic, so-cial and environmental. As a result, the optimalirrigation water resources allocation plans of dif-ferent water types (surface water and groundwa-ter) are obtained. The proposed multi-objectivemodel is unique by considering water-savingmeasures, irrigation water quality impact factorsand groundwater exploitable quantity in the ir-rigation water resources optimal allocation sys-tem. The obtained results are valuable for sup-porting the adjustment of the existing irrigationpatterns and identify a desired water-allocationplan for irrigation.

4. A logística humanitária e a logística em-presarial: relações e desafios.

Sandra Silva (sandrasilva@esa.ipvc.pt) Insti-tuto Politénico de Viana do Castelo e INESCCoimbra, Sandra Silva

Keywords: Logística humanitária, Logística em-

presarial, Processos logísticos, Apoio à decisão multi-

critério e multiobjectivo.

O presente trabalho busca comparar e interligara logística humanitária e a logística empresarial,identificando os desafios a serem enfrentadospela logística humanitária, nos quais a logísticaempresarial pode apoiar na adoção de medidasde desempenho e de melhoramento, criando umcaminho de sinergia e integração entre ambase proximidade académica. O trabalho começapor abordar os conceitos, contrastes e com-parações entre o enfoque logístico empresariale o humanitário. De seguida, apresentam-se ecomparam-se os processos logísticos de uma in-stituição de cariz solidário e uma empresa, am-bas da área alimentar, com missões distintas. Noentanto, com afinidades comuns nos processoslogísticos, como a recolha, o armazenamento ea distribuição e ainda, na necessidade de tomardecisões quanto à localização de infraestruturase à otimização de processos. Por fim, será re-forçado o desafio e o papel dos modelos mul-ticritério e multiobjectivo como ferramentas noapoio à decisão na definição dos processos logís-

ticos humanitários tal como nos empresariais.

Thu.1.4, Thursday, June 29, 9:30-10:50, Room4

Health and Life Science ISession chair: Filipe Alvelos

1. Economies of scope in health sector: thecase of Portuguese hospitals

Diogo Cunha Ferreira(diogo.cunha.ferreira@tecnico.ulisboa.pt)CESUR, CERIS, Instituto Superior Técnico,University of Lisbon, Diogo Cunha Ferreira,Rui Cunha Marques, Alexandre Morais Nunes

Keywords: Hospitals; Economies of Scope; Na-

tional Health Service; Efficiency; Partial Frontier Non-

parametric Methods; Linearization

Background: Economies of scope are defined asthe potential cost savings arising from the jointproduction of two or more outputs, rather thantheir separate production. Given the importanceof the health sector for the community, mea-suring the existence of potential economies ofscope clearly contributes to the improvement ofthis sector sustainability. Objective(s): To searchfor the existence of economies of scope in Por-tuguese hospitals. Methods: This paper devel-ops (1) a generalized algorithm to get locallyconvex frontiers from the bidirectional condi-tional order-alpha frontier method, as well as (2)a generalized economies of scope-based ratio, tobe possible the introduction of all inefficiencysources. Data: This paper makes use of the 2002-2009 Portuguese hospitals’ dataset, concerningthe provision of obstetrics & gynecology & pe-diatrics and psychiatric services. Results: Con-siderable economies and diseconomies of scopewere found in the Portuguese public hospitals. Astrong dependence on the production line andon the merging status of the hospital was ob-served. Diseconomies of scope are more likelyfor large hospitals. Conclusions: Even mergedhospitals can exploit economies of scope; how-ever, when they become outsized entities, suchan exercise turns out to be a hard task.

32 IO 2017

SESSIONS

2. What future for healthcare expenditures?A time series analysisSofia Cruz-Gomes (up200401798@fe.up.pt)FEUP, Sofia Cruz-Gomes, Mário Lopes,Bernardo Almada-Lobo

Keywords: Healthcare expenditure, gross domes-

tic product, healthcare workforce, demographics, time

series analysis.

The healthcare sector has a major relevance inboth social and economic contexts. The ris-ing healthcare expenditures experienced overthe last decades have urged researchers to re-view the determinants of healthcare expendi-tures. The main goals of this study are to assessthe main determinants of health expenditures,to provide a comprehensive understanding onthe relationship between expenditures and someof their main determinants and to predict the fu-ture healthcare expenditures. Using data fromPortugal for the period 1970-2014, we carry outa time series analysis to detect and describe therelationship between healthcare expenditures,economic growth, healthcare workforce and de-mographic variables. Using projections for theGross Domestic Product, the size and age of thepopulation, and the number of Health HumanResources, we project healthcare expendituresuntil 2060. Results show expenditure in health-care depends on the economic conditions, thedemographic structure of the population andon the level of health workforce. The projec-tion of future expenditures based on our resultsand on what is the expected evolution of thosedrivers reveals a rising trend in health expen-ditures, which may generate heightened chal-lenges to governments and managers on the fu-ture delivery of healthcare.

3. A novel participatory approach to scenariobuilding: application to the evolution of pop-ulation health inequalities in EuropeLiliana Freitas (lil-iana.freitas@tecnico.ulisboa.pt) Centrode Estudos de Gestão do Instituto SuperiorTécnico, Universidade de Lisboa, AntónioAlvarenga, Ana Vieira, Liliana Freitas, MónicaOliveira, Carlos Bana e Costa

Keywords: Scenarios; Delphi; scenario building;

scenario planning; online platform; uncertainty; fore-

sight

Acknowledgement: The authors would like to ac-

knowledge the support provided by EURO-HEALTHY

project that has received funding from the European

Union’s Horizon 2020 research and innovation pro-

gramme under Grant Agreement No 643398.

Scenarios are key to capture uncertainty and in-form strategy and policy evaluation. Recognisedcross-cutting challenges in scenario building arethe involvement of a large and diverse number ofparticipants and the combination of their opin-ions with evidence. We present a new approachfor scenario building that entails a sequence offace-to-face and non-face-to-face participatoryprocesses: a two-round Delphi process with alarge panel of experts and stakeholders describ-ing, in a first idea-generation round, causes forfuture changes, which are subsequently workedinto drivers that are then assessed by partici-pants in a second Delphi round; next, in a two-workshop process with a strategic small group,drivers are organized into scenario structures re-inforced by future-oriented evidence, for whichscenario narratives are developed. Technically,a protocol is devised to convert causes intodrivers; and the Extreme-World method is usedto build extreme but plausible scenarios, witha third interim scenario being set. We showhow this approach was applied in the H2020EURO-HEALTHY project to develop scenarios toappraise the evolution of population health in-equalities in Europe.

4. A priori crossmatching in kidney exchangeprogramsFilipe Alvelos (falvelos@dps.uminho.pt) Uni-versidade do Minho, Filipe Alvelos, Ana Viana

Keywords: Kidney excgange program, integer

programming

Kidney exchange programs rely on the definitionof pools of incompatible donor-patient pairs andon the information about the compatibilities ofdonor and patients of different pairs. Assumingsome objective, e.g. maximizing the number oftransplants, a set of transplants can be plannedsuch that the donor of the pair of a patient re-

IO 2017 33

SESSIONS

ceiving a kidney also donates. Commonly, af-ter a transplants plan is defined, actual cross-matching tests are performed for each pair in-volved in a planned transplant. If the result ofthe crossmatching test is positive, the plannedtransplant must be cancelled, as well as the oth-ers transplants involved in the exchange. Wediscuss a different approach: to conduct actualcrossmatching tests before a plan is decided. Be-fore a plan is decided, we select a potential trans-plant and conduct the actual crossmatching test.Based on its result, another potential transplantis selected and the same procedure is repeated.Since conducting crossmatching tests involvestime and money, the approach is limited by amaximum number of potential transplants thatcan be tested. We propose two different meth-ods to decide which crossmatching tests to con-duct and compare their results with the standardprocedure.

Thu.1.5, Thursday, June 29, 9:30-10:50, Room5

Discrete Optimization, Graphs,and Geometry ISession chair: Domingos Cardoso

1. Existence of Nash equilibria on integerprogramming gamesMargarida Carvalho(m@margaridacarvalho.org) École Poly-technique de Montréal, Margarida Carvalho,Andrea Lodi, João Pedro Pedroso

Keywords: Integer programming games; Nash

equilibria; Computational Complexity

Acknowledgement: The first author acknowl-

edges the support of the Portuguese Foundation for

Science and Technology (FCT) through a PhD grant

number SFRH/BD/79201/2011 (POPH/FSE program),

the ERDF European Regional Development Fund

through the Operational Programme for Competitive-

ness and Internationalisation - COMPETE 2020 Pro-

gramme within project POCI-01-0145-FEDER-006961,

and National Funds through the FCT (Portuguese

Foundation for Science and Technology) as part of

project UID/EEA/50014/2013.

We aim to investigate a new class of games,

where each player’s set of strategies is a unionof polyhedra. These are called integer program-ming games. We prove that it is a Σ

p2 -complete

problem to decide the existence of Nash equilib-ria and we provide sufficient conditions for anequilibrium to exist. Additionally, we describeexamples suitable to be modeled by an integerprogramming game: the knapsack game, thecompetitive lot-sizing and the kidney exchangegame.

2. Planning wood chips delivery to bioenergycentralsRicardo Soares (ricardo.f.soares@inesctec.pt)INESC TEC, Ricardo Soares, Alexandra Mar-ques

Keywords: synchronization, vehicle routing prob-

lem, pickup and delivery, mixed-integer programming

Acknowledgement: This work is financed by

the ERDF - European Regional Development Fund

through the Operational Programme for Competitive-

ness and Internationalisation - COMPETE 2020 Pro-

gramme and by National Funds through the Por-

tuguese funding agency, FCT - Fundação para a

Ciência e a Tecnologia within project POCI-01-0145-

FEDER-016733.

This research addresses a wood chips deliveryproblem in the biomass sector by integratingwood-chipping operations on residue piles’ lo-cations and its subsequent transportation. Plan-ning of wood chips’ transportation to bioenergycentrals is integrated with chipping machines’transportation to residue piles, therefore arisingthe need for inter-vehicle and operations syn-chronization in order to complete all operationsin the shortest time possible. At residue piles,chipping machines process and load the mate-rial directly into chips-moving trucks, so they arerequired to be present at the pile during the en-tire chipping process. Furthermore, as chippingmachines are non-autonomous vehicles, theyneed to be transported between residue pilesby auxiliary vehicles. The proposed approachconsists in a Mixed Integer Programming modelwhere three vehicle routing problems are for-mulated and intertwined with synchronizationconstraints. The model also takes advantage ofcomplex networks design in order to embed syn-

34 IO 2017

SESSIONS

chronization aspects and significantly improveits scalability. A solution method approach forthis problem is also presented.

3. A decomposition approach for the long-term scheduling of a single-source multi-product pipeline networkWilliam Hitoshi Tsunoda Meira(williammeira@alunos.utfpr.edu.br) GraduateProgram in Electrical and Computer Engi-neering, Federal University of Technology -Paraná, William Hitoshi Tsunoda Meira, Le-andro Magatão, Ana Barbosa-Póvoa, SusanaRelvas

Keywords: multiproduct pipeline, scheduling, de-

composition approach, mixed integer linear program-

ming

This paper proposes a decomposition approachcombining heuristic algorithms and Mixed In-teger Linear Programming (MILP) models tosolve the long-term scheduling of a multiprod-uct pipeline connecting a single-source to mul-tiple distribution centers. The solution consid-ers many operational aspects, such as simulta-neous deliveries, pipeline maintenance periods,deliveries of multiple products during the samepumping run, and rigorous inventory control.To validate the proposed approach a long-termscheduling of a problem from the literature issolved. This problem is composed of a straightpipeline connecting a refinery to 3 distributioncenters and transporting 4 different oil deriva-tives. The approach was able to obtain an opera-tional solution in less than half a minute of CPUtime.

4. Grafos com número de estabilidadequadrático convexo: principais resultados eproblemas em abertoDomingos Cardoso (dcardoso@ua.pt) Univer-sidade de Aveiro, Domingos Cardoso

Keywords: Número de estabilidade de grafos, op-

timização combinatória, programação convexa

A determinação do número de estabilidade deum grafo é um problema NP-completo. No en-tanto, existem grafos G cujo número de esta-bilidade é determinado pela resolução de umprograma quadrático convexo P(G) definido à

custa da sua matriz de adjacência e menor valorpróprio. Estes grafos, conhecidos por grafoscom número de estabilidade quadrático con-vexo, são também designados por Q-grafos. Oseu estudo inicou-se na década de 90, tendo-se desde logo obtido uma condição necessáriae suficiente para um grafo ser um Q-grafo. Noentanto, esta condição não é computacional-mente efectiva, no sentido em que conduza aum algoritmo polinomial para o seu reconhec-imento. Em 2001, surgiu um algoritmo polino-mial que reconhece se um grafo é ou não umQ-grafo ou devolve um subgrafo H designadopor adverso (grafo para o qual o valor óptimo deP(H) é inteiro, permanecendo constante quandose eliminam os vizinhos de qualquer vértice eo mesmo acontecendo relativamente ao menorvalor próprio). A investigação actual sobre o re-conhecimento de Q-grafos incide precisamentesobre estes grafos adversos. Nesta apresentação,faz-se uma descrição da pesquisa até agora efec-tuada, enfatizando-se os principais resultadosobtidos e os problemas e conjecturas que semantêm em aberto.

Thu.1.6, Thursday, June 29, 9:30-10:50, Room6

Location, Logistics, Transporta-tion, and Traffic ISession chair: Ana Moura

1. Energy minimizing p-connected coveringlocation problem

Arsham Atashi Khoei (ar-sham.khoei@metu.edu.tr) Middle EastTechnical University, Arsham Atashi Khoei,Mohammad Saleh Farham, Mustafa KemalTural

Keywords: Connected coverage problem, Contin-

uous location problem, Second order cone program-

ming

In this study, we consider the energy minimizingp-connected covering location problem (EPC-CLP). Given a set of weighted demand points tobe covered, the EPCCLP aims to find the loca-tions of p connected facilities in the plane in or-der to minimize the total weighted energy con-

IO 2017 35

SESSIONS

sumption of the demand points. The energyconsumption of each demand point is propor-tional to the square of its distance from the near-est facility. We assume that two facilities arejoined by a virtual edge if the distance betweenthem is less than or equal to a given connectivityradius. A set of facilities are said to be connectedif the graph formed by the facilities and all possi-ble virtual edges is a connected graph. We showthat the EPCCLP can be formulated as a mixedinteger second order cone programming prob-lem. For a set of problem instances, the EPC-CLP is solved using a commercial solver. To beable to solve larger size instances of the EPCCLP,a heuristic based on the problem characteristicsis developed. The proposed heuristic is com-pared with the exact solution approach in termsof computational time and solution quality.

2. An intercontinental replenishment prob-lem: A hybrid approach

Elsa Silva (emsilva@inesctec.pt) INESC TEC,Elsa Silva, António Ramos, Manuel Lopes, Pa-trícia Magalhães, José Fernando Oliveira

Keywords: Replenishment Problem, Container

Loading Problem, Hybrid Algorithm

This work addresses a case study in an inter-continental supply chain. The problem emergesin a company in Angola dedicated to the tradeof consumable goods for construction build-ing and industrial maintenance. The companysends the replenishment needs to a Portuguesecompany, which takes the decision of whichproducts and in which quantities will be sent byshipping container. The decision of which prod-ucts and in which quantity should take into con-sideration a set of practical constraints: the max-imum weight and volume of the cargo and finan-cial constraints related with the minimum valueof purchases that guarantees the profitabilityof the business and a maximum value associ-ated with shipping insurance. A 2-stage hybridmethod is proposed. In the first stage, an integerlinear programming model is used to select theproducts that maximise the sales potential. Inthe second stage, a Container Loading Algorithmis used to effectively pack the selected productsin the shipping container ensuring the geomet-

rical constraints, and safety constraints such asweight limit and stability. A new set of probleminstances was generated with the 2DCPackGenproblem generator, using as inputs the data col-lected in the company. Computational resultsfor the algorithm are presented and discussed.

3. Otimização de rotas para abastecimentode linhas de montagemJosé Correia (jmmcorreia94@ua.pt) Universi-dade de Aveiro, José Correia, Ana Moura

Keywords: Planeamento de rotas para veículos,

Logística interna, programação linear inteira, métodos

heurísticos.

Este trabalho foi desenvolvido para uma em-presa Portuguesa, com o intuito de otimizaras rotas diárias dos comboios logísticos(mizusumashi). O objetivo principal passa porreduzir o tempo do processo de abastecimentodas linhas de montagem do produto final, como objetivo de reduzir custos de energia e demão-de-obra. De acordo com as necessidadesde produção, são definidas diariamente rotaspara os diversos mizusumashis, que entregame recolhem material em pontos pré-definidosdas linhas. Para isso, as rotas têm de ser calculasde acordo com: necessidades produtivas, ospontos de carga e descarga, os tempos deserviço em cada ponto, os tempos de deslocaçãoentre os pontos e precedências na ordem devisita dos mesmos. Foi então desenvolvidauma abordagem que recorre a um modelo deprogramação linear inteira, para cálculo dasrotas diárias. Deste modo o operador é capazde adaptar as diferentes rotas existentes às maisdiversas necessidades diárias da fábrica. Paravalidação da abordagem desenvolvida, foramtestados vários cenários possíveis, recorrendosempre a dados reais e concretos da empresa.Estes resultados obtidos são posteriormentecomparados com os dados históricos da em-presa, verificando-se uma melhoria significativaem termos de redução de tempos totais dedeslocação dos mizusumashi.

4. Otimização da gestão de agendamento deprestação de cuidados de saúdeAna Moura (ana.moura@ua.pt) CIDMA - Uni-versidade de Aveiro, Jorge Azevedo, Ana Moura

36 IO 2017

SESSIONS

Keywords: Problemas de afetação, Programação

linear inteira, Abordagens heurísticas

Este trabalho foi desenvolvido para responder àsnecessidades de um agrupamento de Unidadesde Saúde, que dá apoio domiciliário prestandocuidados de saúde primários. O agrupamentotem atualmente sete viaturas, mas só em qua-tro é que existe controlo efetivo do seu agen-damento às prestações de cuidados de saúdediárias. Esse agendamento é feito em papel,sendo a afetação das viaturas efetuada de umaforma ad hoc, de acordo com a ordem dechegada dos pedidos via e-mail e com a priori-dade de algumas das unidades de saúde. Alémdisso, não existe histórico fidedigno dos serviçosprestados diariamente e dos respetivos carrosutilizados, assim como dos quilómetros efetu-ados. A ideia deste trabalho passa, não só porotimizar a gestão da frota existente nas unidadesde saúde, como também manter um históricosobre: os serviços domiciliários efetuados; osveículos utilizados; e respetivos quilómetros to-tais percorridos diariamente. Para isso, foi de-senvolvido um algoritmo híbrido que integra ummodelo de programação linear inteira para aafetação dos veículos às prestações de serviço,com a definição dos diversos tipos de serviços aprestar diariamente e respetiva priorização. Estaabordagem foi testada com dados reais forneci-dos pelo referido agrupamento de Unidades deSaúde.

Thu.2.1, Thursday, June 29, 11:20-13:00, Room1

Production Management and Sup-ply ChainSession chair: Ana Raquel Xambre

1. The time window assignment vehicle rout-ing problem with product dependent deliver-ies

Fábio Neves-Moreira(fabio.s.moreira@inesctec.pt) Faculty ofEngineering of the University of Porto andINESC TEC, Fábio Neves-Moreira, DiogoPereira da Silva, Luís Guimarães, BernardoAlmada-Lobo

Keywords: Time Window Assignment, Vehicle

Routing, Integrated Planning, Fix And Optimize, Real-

World Application

The Time-Window Assignment Vehicle RoutingProblem (TWAVRP) arises in a context wherecustomers need to stop their core activities forreceiving goods. Time-windows are usually de-fined considering the trade-off between cus-tomers’ constraints and transportation planningefficiency. We study a challenge inspired bya Portuguese retailer holding two warehousesand a fleet of vehicles to serve orders posted bystores. Hard business constraints enforce thestores to demand certain types of products (i.e.fresh, frozen) at very specific periods of the day.Consequently, some time periods have manyoverlapping time-windows. Peaks on the needfor vehicles may be triggered if time-windowsare not strategically assigned to disperse deliv-eries along the day. We propose a novel math-ematical formulation to redefine product-typedependent time-windows, complying with busi-ness constraints. Representative demand sce-narios are considered to simultaneously definedelivery schedules and time-windows which areto remain constant after the optimization pro-cess. The goal is to satisfy demand while mini-mizing the costs associated to the number of ve-hicles and travelled distance. The value of theproduct-type dependent TWAVRP is assessedby means of real-world instances. A sensitiv-ity analysis on the number of time-windowschanged, where quick-wins are obtained by in-troducing small changes on the retailer’s solu-tion.

2. A matheuristic for the irregular strip pack-ing problem

Larissa Oliveira (ltdo@icmc.usp.br) Instituteof Mathematics and Computer Science,University of São Paulo, Larissa Oliveira,Franklina Toledo, Maria Antónia Carravilla,José Fernando Oliveira

Keywords: Strip packing problem; Matheuristic

The cutting and packing (C&P) problem consistsin cutting a set of small pieces from a large ob-ject minimizing the raw-material waste. In this

IO 2017 37

SESSIONS

study, we focus on the irregular strip packingproblem, one of the many variants of the C&Pproblem. In the irregular strip packing problem,the small pieces have irregular shapes and thegoal is to minimize the used length of a large ob-ject with fixed height. To tackle this problem,a matheuristic based on the biased random-keygenetic algorithm (BRKGA) is proposed. The de-coder is a linear programming model that min-imizes the used length and its non-overlappingconstraints are written based on the edges ofthe nofit polygons (NFP) of the pairs of pieces.Each solution is encoded as a vector of n ran-dom keys, where n is the total number of NFP.The keys encode which edges of the NFP will beused to guarantee the non-overlapping betweenthe pieces.

3. Postponement: definição do ponto de de-sacoplamento com recurso a um procedi-mento para a formação de famílias de produ-tos

Ana Raquel Xambre (raquelx@ua.pt)DEGEIT/CIDMA - Universidade de Aveiro,Ana Raquel Xambre

Keywords: postponement, ponto de desacopla-

mento, famílias de produtos

A estratégia de postponement (ou diferenci-ação tardia do produto) é, cada vez mais, umaopção nas empresas industriais, pois permitecombinar a lógica existente num ambiente MTS(make-to-stock) com as ideias presentes numsistema MTO (make-to-order). Assim, o iní-cio do processo produtivo é estandardizado eidêntico para um elevado número de compo-nentes (similar à lógica MTS), mas a parte finaldo processo inclui um conjunto de operaçõesque permitem diferenciar os produtos, incorpo-rando nestes os requisitos do cliente (similar auma lógica MTO). O ponto de desacoplamentocorresponde ao local (operação) do processo defabrico em que é iniciada a fase de diferenciaçãodo produto, sendo crítica a definição adequadadeste, quando se pretende aplicar uma estraté-gia de postponement. Neste trabalho é apre-sentado um procedimento, definido com baseem algoritmos genéticos, que permite definirfamílias de produtos idênticos. A construção das

famílias é feita considerando a semelhança dosprocessos de fabrico dos diferentes produtos, oque facilita a definição dos pontos de desacopla-mento para cada família.

Thu.2.2, Thursday, June 29, 11:20-13:00, Room2

Multicriteria IISession chair: Monica Oliveira

1. Mapping the use of multicriteria methodsapplied to public securityMarcio Basilio (marciopbasilio@gmail.com)Federal Fluminense University, Marcio Basilio,Helder Costa, Valdecy Pereira

Keywords: Webibliomining, public security, mul-

ticriteria analysis

Acknowledgement: We thank the Federal Flumi-

nense University (UFF) and Coordination of Improve-

ment of Higher Level Personnel (CAPES) for the sup-

port of the research.

This work is based on the assumption that fewstudies have been conducted in the public se-curity sector employing multiple criteria tech-niques to support the decision making proce-dure. In this context, this research had the objec-tive of mapping the application of MCDA meth-ods within the scope of public security. A sys-tematic search was conducted by applying theWebibliomining method. After applying specifickeywords and filters 965 records were returnedfrom Scopus and 788 records from the ISI Webof Science. The review of findings summary re-sulted in a set of 19 articles validated as com-pliant for inclusion in the revision, where 19were from Scopus and 6 were shared with theWeb of Science base. The analysis of the articlesshowed that there is no predominant multicrite-ria method in the context of the search. The re-current applications refer to the resolution of thepolicing distribution problem, and classificationof crime rate by region. The key contributionof this work is to provide the reader with a mapof public safety problems already addressed inthe context of MCDAs approaches and to high-light aspects of this problem that have not yet ex-plored in the context of the MCDA.

38 IO 2017

SESSIONS

2. Métodos de decisão multicritérios: um es-tudo empírico aplicado à segurança públicaMax Oliveira (prof.maxadm@gmail.com) Uni-versidade Federal Fluminense, Marcio Basilio,Helder Costa, Valdecy Pereira, Max Oliveira

Keywords: Multicritério, PROMETHEE, Produtivi-

dade, Ordenação, Segurança Pública

Acknowledgement: Agradecemos à Universidade

Federal Fluminense - UFF e Coordenação de Aper-

feiçoamento de Pessoal de Nivel Superior (CAPES) pelo

apoio a pesquisa.

O presente estudo objetivou efetuar compara-ção entre o método utilizado no sistema inte-grado de metas para criação de rank entre asÁreas Integradas de Segurança Pública - AISP,para premiação por produtividade no Estado doRio de Janeiro e o MCDA para um problema rela-tivo a ordenação denominado Preference Rank-ing Organization Method for Enrichment Eval-uations - PROMETHEE. Utilizou-se para efeitosdesta pesquisa, a base de dados da 12ª. ediçãodo sistema integrado de metas relativo premi-ação do 1º. semestre de 2015. Esta edição foi es-colhida, pois foi a que apresentou maior númerode AISPs empatadas em primeiro lugar, tendo ocusto de todo processo ter sido superior a 50%do somatório das onze edições anteriores. O re-sultado do método PROMETHE II, apresentoualteração de posição em mais de 85% da clas-sificação total, com alterações significativas en-tre as quatro primeiras posições, com alteraçãode posições e inclusões neste conjunto, sem em-pates e com classificações distintas. Aplicaçãodo PROMETHE II mostrou-se mais eficiente emrelação ao Sistema Integrado de Metas - SIM,pois além de eliminar o efeito compensatórioque afetam os cálculos baseados em médias,proporcionou um rank mais bem definido o queimpactaria o programa com redução de custo.

3. Shaping Delphi processes to inform mul-ticriteria decision analysis in the healthcarecontextAna Vieira(ana.lopes.vieira@tecnico.ulisboa.pt) In-stituto Superior Técnico, Universidade deLisboa (IST-UL), Ana Vieira, Carlos Bana eCosta, Mónica Oliveira, João Bana e Costa

Keywords: Delphi; online platform; multicriteria

decision analysis; MACBETH; participatory methods

Acknowledgement: The authors would like to ac-

knowledge the support provided by EURO-HEALTHY

project that has received funding from the European

Union’s Horizon 2020 research and innovation pro-

gramme under Grant Agreement No 643398.

Multicriteria Decision Analysis provides soundmethods used in many complex decision con-texts to develop robust evaluation models. Fur-ther to technical meaningfulness, there is alsoa requirement for the design of adequate socialprocesses to capture the points of views of mul-tiple actors involved. This study explores the useof the Delphi method as a way to collect and in-corporate a large number of experts’ and stake-holders’ judgments in multicriteria evaluationmodels in health, and focuses on the design andimplementation of Delphi to inform the con-struction of a Population Health Index (PHI) in areal context. We detail a Web-Delphi for weight-ing, where MACBETH qualitative value judg-ments were elicited; and a Web-Delphi for valuefunctions, where the shape of the value functionon each indicator was determined. Both wereimplemented through the Welphi technologicalplatform and their qualitative outputs reached ahigh level of agreement; stability and reliabilityanalyses enabled validation of the results; andDelphi results guided a strategic group in theprocess of building an evaluation model in a de-cision conference setting. In conclusion it is dis-cussed to which extent the Delphi method canbe a knowledge acquisition and interchange toolin different decision-aiding contexts in health.

4. Overcoming common critical mistakesin building composite indices through theMACBETH socio-technical approach

Monica Oliveira (mon-ica.oliveira@tecnico.ulisboa.pt) IST -Universidade de Lisboa, Monica Oliveira,Teresa Rodrigues, João Bana e Costa, AnaVieira, Ãngela Freitas, Liliana Freitas, PaulaSantana, Carlos Bana e Costa

Keywords: multidimensional indices; compos-

ite indices; multi-criteria value measurement; MAC-

IO 2017 39

SESSIONS

BETH; socio-technical approach

Composite indices have been used as formal as-sessment tools, but are often built with ad hocprocesses and suffer from theoretical meaning-fulness. Departing from common critical mis-takes of composite indices, in this presentationwe show how these can be avoided by follow-ing the principles of multi-criteria value mod-elling. Specifically we describe how the MAC-BETH socio-technical approach has been de-vised to overcome these problems and to assistpopulation health indices construction in thecontext of the EURO-HEALTHY H2020 and ofthe GeohealthS FCT projects. Under this techni-cal framework, the added value to health of im-proving performance on a criterion is measuredby a value-function (or by a conditional value-function in case of interdependencies) and therelative importance of performance improve-ments in different criteria is measured by relativeweights. The judgemental information to de-fine value-functions and weights can be elicitedwith MACBETH through two-stage participatoryprocesses: a Delphi panel formed by a largenumber of experts and stakeholders to agreeupon value-functions shapes and "qualitative"weights, which then informs a decision confer-encing process with a small strategic group. Wedescribe how this approach was successfully ap-plied to build indices at the municipality level inPortugal and at the regional level in Europe.

Thu.2.3, Thursday, June 29, 11:20-13:00, Room3

Energy, Environment, Natural Re-sources, and Weather ISession chair: Amilcar Arantes

1. OR methods applied to forest-based supplychains considering economic, environmen-tal and social perspectives - A systematic re-view

Andreia Santos (an-dreia.d.santos@tecnico.ulisboa.pt) InstitutoSuperior Técnico, Andreia Santos, Ana Póvoa,Ana Carvalho

Keywords: Operational Research, Forest, Supply

Chain, Sustainability

The raising public awareness on sustainabil-ity has obliged companies to consider in theirdecision-making process the three sustainabilitypillars (economic, environmental and social). Inthis context the forest sector is not an exception.Thus the goal of the present work is, througha systematic literature review, identify what ORbased-methods have been used in the study offorest supply chains considering the three pil-lars of sustainability within the different decisionlevels (strategic, tactical or operational) . Sci-enceDirect was the database used and a time pe-riod from 1995 to 2016 was analyzed. A total of 62papers form the final sample. Optimization wasidentified as the most used OR method. Otherwidely used techniques include heuristics, sim-ulation and multi-criteria analysis. Within thesemethods the most treated sustainability criteriaare: costs (economic); greenhouse gas emissions(environmental); and job creation (social). Onthe decision levels the strategic one has beenthe most addressed and uncertainty has beenmainly considered in raw materials supply, finalproducts demand and prices. As main conclu-sion, it can be stated that much more needs tobe done to integrate the three pillars of sustain-ability in the decision-making process regardingforest-based supply chains.

2. Downstream petroleum supply chains’design and planning - contributions androadmap

Leão José Fernandes (leao.fernandes@clc.pt)CLC Companhia Logística de Combustíveis,EN 366, Km 18, 2050-125 Aveiras de Cima,Leão José Fernandes, Susana Relvas, Ana PaulaBarbosa-Póvoa

Keywords: Petroleum supply chain, Strategic and

tactical planning, Uncertainty and risk management,

Stochastic MILP, Robust optimization

Petroleum Supply Chains (PSC) networks arecomplex organizations, strongly affected bycompetition, environmental regulation andmarket uncertainty. To improve profits andreduce costs and risks, companies may usemathematical programming for strategic, tac-

40 IO 2017

SESSIONS

tical and operational planning. The currentpaper identifies the research opportunities andpresents our contributions with respect to thestrategic and tactical planning of multiple entity,echelon, product, and transportation PSCs,under the context of crude costs, product prices,and customer demand uncertainties. In orderto address these gaps, four mixed integer linearprogramming (MILP) models were developed,namely the individualistic, collaborative, mul-tiobjective stochastic, and robust optimizationMILPs. A detailed pricing structure and apiecewise linearization function determine thecollaborative economy of scale multi-entitycosts, tariffs and prices per route, locationand product. A stochastic programming MILPintegrates an augmented e-constraint algorithmto simultaneously maximize the expected netpresent value (ENPV) and minimize risk repre-sented through selected measures. The robustoptimization MILP optimizes the worst-caseprofits considering the crude costs, productprices, and customer demand uncertainties.Test results are presented for the Portuguesedownstream PSC.

3. A collaborative platform to enhance sus-tainability in forest-based supply chainsAlexandra Marques (asfmar-ques78@gmail.com) , Alexandra Marques,Pedro Amorim, Ana Carvalho, Jorge Pinho deSousa, Ana Póvoa

Keywords: forest-based supply chains, life cycle

assessment, simulation-optimizaton, logistics, collab-

oration

Acknowledgement: This work is financed by

the ERDF - European Regional Development Fund

through the Operational Programme for Competitive-

ness and Internationalisation - COMPETE 2020 Pro-

gramme and by National Funds through the Por-

tuguese funding agency, FCT - Fundação para a

Ciência e a Tecnologia within project POCI-01-0145-

FEDER-016733.

The aims of higher efficiency and sustainabilityforest-based industries poses three main chal-lenges. The first - Supply Chain SustainabilityAssessment - is characterizing resource-efficientsupply chains and how to assess the sustainabil-

ity dimensions - social, economic and environ-mental in forest-based supply chains. The sec-ond - Supply Chain Planning - is how to opti-mally plan the efficient use of the transportationresources distributed among haulers, minimiz-ing the ecological impact of operations and ful-filling the due dates settled by the industry, whileaccounting for the uncertainty of real-life oper-ations. The third - Supply Chain Collaboration- is how to set into practice the new processesand tools for resource-efficient supply chainsin a collaborative environment at the companylevel, and at the supply chain level. Over theyears, significant research has been done inthese domains. However, most of the researchefforts have followed separated ways.The pro-posed platform will combine Life Cycle Assess-ment methods, simulation-optimization meth-ods for logistics planning and collaborative in-centives. It will be tested in the wood-based pan-els industry in Portugal. Preliminary results of itsapplication will be presented.

4. Towards sustainability in the constructionand demolition waste industry: a case studyon application of lean methodology

Amilcar Arantes (amil-car.arantes@tecnico.ulisboa.pt) CERIS,CESUR, Instituto Superior Técnico, Univer-sidade de Lisboa, Mariana Ribeiro, MartaCastilho Gomes, Amilcar Arantes

Keywords: Construction and Demolition Waste,

Lean, Operations Management, Sustainability

The construction industry is one of the largestconsumers of natural resources and one of themain generators of construction and demolitionwaste (CDW). Thus, the incorporation of recy-cled aggregates in construction is highly rele-vant to achieve sustainability in the construc-tion industry. However, recycling of CDW is onlyfeasible if aggregates are competitive in relationto new construction materials. This work fo-cuses on a Portuguese company that managesthe CDW generated by its demolition works, andaims at increasing the efficiency of the produc-tion process of recycled aggregates. The mainobjective is to propose the implementation ofimprovement actions based on lean manage-

IO 2017 41

SESSIONS

ment following the DMAIC cycle (Define - Mea-sure - Analyze - Improve - Control) methodology,an organized and sequential method for solvingproblems using various tools and methodologiesfrom the six sigma management philosophy. Thestudy of recycling operations at the companyled to identify lean wastes, understand theirroot causes and propose improvement strate-gies and respective forms of maintenance. Theproposed improvement measures were designedin the future value stream mapping showing in-creased production capacity, reduction of leadtime and consequently enhanced quality of theaggregates, lower production costs and strength-ening of the company competitiveness.

Thu.2.4, Thursday, June 29, 11:20-13:00, Room4

Health and Life Science IISession chair: Maria Eugénia Captivo

1. Warfarin dose optimization with geneticalgorithms

Susana Rafaela Martins (malea-far@gmail.com) Algoritmi R&D Centre,Susana Rafaela Martins, Lino Costa, PedroOliveira

Keywords: Anticoagulante therapy,optimization,

genetic algorithms

Anticoagulant therapy is a treatment that is ap-plied when it is needed to monitor blood co-agulation levels. Since these levels are difficultto control, it is important to develop modelsthat allow the prediction of the effect of differ-ent anticoagulant doses. In this work, it is pro-posed a model to for warfarin dosage optimiza-tion. The objective is to determine the optimaldosage along time to achieve a desirable bloodcoagulation level. This model includes severalparameters related with patient condition suchas genotype information and age. Several in-stances of the optimization problem are solvedby Genetic Algorithms. The results indicate thatthis approach can be useful in the specificationof the best patient warfarin dosage.

2. Planning health workforce training in thedetection and prevention of excessive alco-

hol consumption: an optimization-based ap-proachJoana Faria (joana.c.a.faria@gmail.com) Cen-tre for Management Studies of Instituto Su-perior Técnico (CEG-IST), Joana Faria, TeresaCardoso-Grilo, Cristina Ribeiro

Keywords: Health workforce training, excessive

consumption of alcohol, optimization, mathematical

programming models, performance indicators

The adequate training of health workforce in thefield of excessive alcohol consumption is essen-tial to provide health professionals with the nec-essary tools for an adequate provision of care,thus leading to a decrease in alcohol consump-tion. Proper planning of such training is thus es-sential, but literature in this area is still scarce.This paper proposes an optimization modelbased on mathematical programming for sup-porting the planning of health workforce train-ing in the field of excessive alcohol consump-tion in National Health Service-based countries- the WFTMalcohol. The model aims at inform-ing on i) how many health professionals (physi-cians and nurses) should be trained per yearand health unit, and ii) which training pack-ages should be available per year. The model al-lows exploring the impact of considering differ-ent objectives relevant in this sector, includingthe minimization of costs and the maximizationof multiple performance indicators. Acknowl-edging that several sources of uncertainty mayaffect planning decisions, a sensitivity analysison key parameters of the model is performed. Toillustrate the applicability of the model, a casestudy based on the Oeste Sul ACES in Lisbon isanalyzed. Results confirm that there is a short-age of trained professionals in this field in Portu-gal.

3. Planning the delivery of long-termcare: Improving location-allocation multi-objective stochastic models through scenarioreduction techniquesTeresa Cardoso-Grilo(teresa.sofia.grilo@iscte.pt) Instituto Uni-versitário de Lisboa (ISCTE-IUL), Centrefor Management Studies of Instituto Supe-rior Técnico, Teresa Cardoso-Grilo, Mónica

42 IO 2017

SESSIONS

Oliveira, Ana Barbosa-Póvoa

Keywords: Long-term care planning, mathemat-

ical programming, stochastic programming, multi-

objective, scenario reduction methods

An adequate planning of long-term care (LTC)delivery currently represents a major challengefaced by European nations. This planning isparticular challenging for countries facing highbudget constraints and with a National HealthService structure in which LTC systems are in-herently planned. Under such LTC contexts itis central to account for multiple policy objec-tives and for the impact of key uncertainties. Ac-cordingly, this study proposes a multi-objectivestochastic mathematical programming model tosupport the planning of LTC networks, in termsof services’ location, capacity planning and pa-tients’ allocation, with emphasis being put onthe stochastic component of the model. Theproposed location-allocation model accountsfor i) multiple health policy objectives, includingequity, cost, health and wellbeing benefits, andii) the impact of uncertainty on planning deci-sions. Given that multiple issues exist regardingthe modelling of uncertainty, scenario reductiontechniques are explored: two scenario reduc-tion techniques proposed in the literature arecompared with a new approach inspired by theforesight literature and that accounts for experts’knowledge within stochastic modelling. A Por-tuguese case study is explored, confirming therelevance for accounting for multiple objectives,as well as for considering new approaches pro-viding more meaningful models and enablingfaster model resolution.

4. Planning healthcare workforce training: amulti-objective mathematical programmingmodel

Marta Monteiro (martacmon-teiro92@gmail.com) Centre for ManagementStudies, Instituto Superior Técnico, Lisbon,Portugal, Marta Monteiro, Teresa Cardoso-Grilo, Mónica Oliveira, Ana Barbosa-Póvoa

Keywords: Health workforce planning, workforce

training, mathematical programming, multi-objective

The planning of health workforce is currently

an important health policy priority across Euro-pean countries, where serious imbalances andinefficiencies in the availability of health hu-man resources have been registered. A properdimensioning of the future healthcare work-force requires a thorough planning of trainingwhich should account for multiple policy ob-jectives, which entail cost, efficiency and eq-uity considerations. Accordingly, this study pro-poses a multi-objective mathematical program-ming model to support the planning of healthworkforce training in countries in which govern-ments regulate and control the training of hu-man resources. Particularly, the model focusesin the training of physicians and informs deci-sions related to the opening of vacancies in med-ical schools and across medical specialties, whileensuring the attainment of multiple objectives- namely, the minimization of oversupply andshortages in the provision of care, the maximiza-tion of equity in the distribution of physiciansacross specialties, and the minimization of costsincurred in the workforce training process. Theaugmented e-constraint method is used to dealwith these multiple objectives. A Portuguesecase study is analysed, showing that substantialadjustments need to be performed on the num-ber of vacancies and on their allocation to spe-cialties in order to anticipate future health sys-tem changes.

5. Local search heuristics for a surgical caseassignment problem

Maria Eugénia Captivo (mecaptivo@fc.ul.pt)Centro de Matemática, Aplicações Funda-mentais e Investigação Operacional, Facul-dade de Ciências, Universidade de Lisboa,Catarina Mateus, Inês Marques, Maria EugéniaCaptivo

Keywords: Health services management, Elective

surgeries scheduling, Heuristics, Local search, Inter-

face

This work is part of an ongoing project with aPortuguese public hospital, where the electivesurgeries scheduling problem is studied. Theadministration of the hospital aims to achievethe targets set by the Portuguese Ministry ofHealth for surgical production and to ensure a

IO 2017 43

SESSIONS

surgical service with a high level of efficiency.However, hospitals do not have an elective surg-eries scheduling system, so the surgeries are un-systematically scheduled and without respectingequity criteria, thus disregarding the Ministry’sgoals. Three versions of the problem are consid-ered and local search heuristics are developed inorder to quickly find good solutions for each ver-sion of the problem. The heuristics select pa-tients from a long waiting list for surgery to bescheduled in each week, and define a room, aday and a shift for each selected surgery accord-ing to one of three different perspectives of theproblem. Tests with real instances are presentedand analyzed. Very good quality solutions withquite low computational time are obtained. Auser-friendly interface was created for the hos-pital.

Thu.2.5, Thursday, June 29, 11:20-13:00, Room5

Nonlinear OptimizationSession chair: Ana Luisa Custodio

1. A penalty approach for solving nonsmoothand nonconvex MINLP problems

M. Fernanda P. Costa (mfc@math.uminho.pt)University of Minho, M. Fernanda P. Costa, AnaMaria A.C. A.C. Rocha, Edite M. G.P. Fernandes

Keywords: MINLP, penalty function, DIRECT, or-

acle

This paper presents a penalty approach for glob-ally solving nonsmooth and nonconvex mixed-integer nonlinear programming (MINLP) prob-lems. Both integrality constraints and generalnonlinear constraints are handled separately byhyperbolic tangent penalty functions. Proximityfrom an iterate to a feasible promising solution isenforced by an oracle penalty term. The numer-ical experiments show that the proposed oracle-based penalty approach is effective in reach-ing the solutions of the MINLP problems and iscompetitive when compared with other strate-gies.

2. A branch and bound method for convexsemi-infinite programming

A.I.F. Vaz (aivaz@dps.uminho.pt) University of

Minho, Hoai An Le Thi, Mohand Ouanes, A.I.F.Vaz

Keywords: Global optimization; Semi-infinite

Programming; Exchange type method; Discretization

scheme; Branch and Bound

Acknowledgement: This work has been sup-

ported by COMPETE: POCI-01-0145-FEDER-007043

and FCT - Fundação para a Ciência e Tecnologia within

the Project Scope: UID/CEC/00319/2013

In this work we propose an efficient method forglobally solving a class of convex semi-infiniteprogramming (SIP) problems. Under convexityassumption of the objective function and con-straints (w.r.t. the variables to be optimized)and appropriate differentiability, we propose abranch and bound type method for SIP. To com-pute a feasible point for a SIP problem (andcheck feasibility) we need to solve a global op-timization (sub-)problem, which is herein ad-dressed by a branch and bound strategy. Themajor novelty of the proposed method consistsin generating a sequence of feasible points forthe SIP problem, obtained by a convex combina-tion of a feasible point and the solution of a dis-cretized finite optimization problem. A branchand bound strategy is also used to address theproblem of minimizing the objective function,since we naturally obtain, as a result of the it-erative process, bounds for the optimum objec-tive function value. Under mild assumptions weprove convergence of the proposed algorithm.To illustrate the proposed method, we providesome numerical results using some benchmarktest problems.

3. A generator of nonregular semidefiniteprogramming problemsTatiana Tchemisova (tatiana@ua.pt) Uni-versity of Aveiro, Eloisa Macedo, TatianaTchemisova

Keywords: semidefinite programming, regularity,

constraint qualification, good behaviour

Regularity plays a key role in the characteriza-tion of optimal solutions in optimization. Thereexist some notions of regularity, such as con-straint qualifications (CQs), well-posedness, andgood behaviour, which are closely related to

44 IO 2017

SESSIONS

each other. Absence of regularity leads to the-oretical and numerical difficulties, and solversmay fail to provide a trustworthy result. There-fore, it is very important to verify if a given prob-lem is regular in terms of some regularity condi-tion and also use specific methods for nonregu-lar problems. In order to test new stopping cri-teria and the computational behaviour of newmethods, having access to sets of reasonably-sized nonregular test problems is a major tool.This paper presents a generator and also adatabase of nonregular Semidefinite Program-ming (SDP) instances with prescribed irregular-ity degrees. Numerical experiments using pop-ular SDP solvers on problems from the newdatabase are carried out.

4. Can optimization help additive manufac-turing?Sérgio Pereira (sergiop@dps.uminho.pt) Uni-versity of Minho, Sérgio Pereira, A. Ismael F. Vaz

Keywords: Additive manufacturing; 3D printing

Acknowledgement: This work is supported by

FIBR3D project - Hybrid processes based on additive

manufacturing of composites with long or short fibers

reinforced thermoplastic matrix, reference SAICT-

PAC/0036/2015, financed by European Structural and

Investment Funds (ESIFs) through the Lisbon Regional

Operational Programme 2020 and by FCT national

funds - Fundação para a Ciência e a Tecnologia, project

POCI-01-0145-FEDER-016414.

Additive manufacturing, also known as layeredmanufacturing and more recently synonymousof 3D printing, has emerged in the last decadesbecoming an alternative to the traditional sub-tractive manufacturing. Contrary from subtrac-tive manufacturing which is a process where 3Dobjects are built by cutting of material from ablock of material, additive manufacturing is aprocess where 3D objects are built by addingmaterial in consecutive layers. This results in aprocess that will require less energy consump-tion and waste of material. However, some lim-itations are pointed out to this process and inparticular to its four stages: part orientation,creation of supports, slicing and path planning.Both orientation and supporting are usually re-lated, since best orientation of the part to be

built can result in lower building time and lesssupport needed, resulting in surface improve-ment. Slicing imply the object division by lay-ers leading to a staircase effect, being more evi-dent for objects with high slopes and curvatures.Path planning consists in the best nozzle path forlayers building. An optimized path planning willavoid the appearance of voids or excess of depo-sition material. This talk will address how opti-mization can help additive manufacturing.

5. An approach based on nonmonotone di-rectional direct search to noisy optimization

Ana Luisa Custodio (alcustodio@fct.unl.pt)Universidade Nova de Lisboa, Ana Luisa Cus-todio, Samuel Marcos

Keywords: derivative-free optimization, noisy op-

timization, directional direct search, nonmonotone

strategies

In industrial applications, mainly in engineer-ing, it is common to be faced with challeng-ing optimization problems where, in particular,the evaluation of the objective function could becontaminated by numerical noise. The presenceof this noise prevents the use of derivative basedoptimization methods. Directional direct searchis one of the classes of optimization methodssuited for this type of optimization. Neverthe-less, when the initialization provided to the opti-mizer is far from the function minimizer or whenin presence of noise with quite irregular oscilla-tions (like is the case of Gaussian noise), mod-ifications are required in order to make it morerobust/efficient. We propose to use nonmono-tone approaches, where the value of the objec-tive function is not required to improve betweenconsecutive iterations, but along an historic ofiterations. This procedure allows to escape fromspurious minima, resulting from the presenceof noise. In this talk we will detail the non-monotonous variants of directional direct searchconsidered, present the corresponding theoret-ical results related to convergence and reporttheir numerical performance.

IO 2017 45

SESSIONS

Thu.2.6, Thursday, June 29, 11:20-13:00, Room6

Location, Logistics, Transporta-tion, and Traffic IISession chair: Maria Isabel Gomes

1. A dynamic programming approach for in-tegrating dynamic pricing and capacity deci-sions in a rental contextBeatriz Brito Oliveira (beat-riz.oliveira@fe.up.pt) INESC TEC and Facultyof Engineering, University of Porto, BeatrizBrito Oliveira, Maria Antónia Carravilla, JoséFernando Oliveira

Keywords: Car rental, dynamic programming, dy-

namic pricing, fleet deployment, optimization model,

constraint programming

Acknowledgement: The first author was sup-

ported by grant SFRH/BD/103362/2014 from FCT -

Fundação para a Ciência e Tecnologia (Portuguese

Foundation for Science and Technology). This

work was also partially financed by the ERDF -

European Regional Development Fund through the

Operational Programme for Competitiveness and

Internationalisation - COMPETE 2020 Programme

within project "POCI-01-0145-FEDER-006961", and

by National Funds through the FCT - Fundação

para a Ciência e Tecnologia (Portuguese Founda-

tion for Science and Technology) as part of project

UID/EEA/50014/2013.

Car rental companies have the ability and po-tential to integrate their dynamic pricing deci-sions with their capacity decisions. Pricing hasa significant impact on demand, while capac-ity, which translates fleet size, acquisition plan-ning and fleet deployment throughout the net-work, can be used to meet this price-sensitivedemand. Dynamic programming has been of-ten used to tackle dynamic pricing problems andalso to deal with similar integrated problems,yet with some significant differences as far asthe inventory depletion and replenishment areconsidered. The goal of this work is to under-stand what makes the car rental problem differ-ent and hinders the application of more com-mon methods. To do so, a discrete dynamic pro-

gramming framework is proposed, with two dif-ferent approaches to calculate the optimal-valuefunction: one based on a Mixed Integer NonLinear Program (MINLP) and one based on aConstraint Programming (CP) model. These twoapproaches are analyzed and relevant insightsare derived regarding the (in)ability of discretedynamic programming to effectively tackle thisproblem within a rental context when realisti-cally sized instances are considered.

2. Cargo stability in the container loadingproblem - State-of-the-art and future re-search directionsAntónio Ramos (agr@isep.ipp.pt) ISEP, AntónioRamos, José Oliveira

Keywords: Static Stability, Dynamic Stability,

Conteiner Loading Problem

The purpose of this paper is to present the cur-rent understanding and conceptualisation of thecargo stability constraint within the context ofthe Container Loading Problem. This problemis highly relevant in the transportation industrydue to the increasing pressure for a more eco-nomically, environmentally and socially efficientand sustainable cargo transportation. Stability isone the most important practical relevant con-straints in the Container Loading Problem dueto its strong influence on the cargo arrangement.Stability is usually divided into stability duringloading operations (static) and stability duringtransportation (dynamic). Two main contribu-tions are made. Firstly, an overview of recent de-velopments in the literature on the two types ofstability, static and dynamic, is provided. Sec-ondly, of opportunities for future research areidentified.

3. Rotas para recolha de valores nos arcosLeonor S. Pinto (lspinto@iseg.ulisboa.pt)Universidade de Lisboa, Instituto Superiorde Economia e Gestão, Lisboa, Portugal,CEMAPRE, Leonor S. Pinto, Miguel Con-stantino, M. Cândida Mourão

Keywords: Rotas nos arcos, formulação de pro-

gramação inteira, mataheurística

Acknowledgement: Os autores agradecem à

Fundação para a Ciência e Tecnologia (FCT). Este

trabalho é parcialmente financiado por FCT/MEC

46 IO 2017

SESSIONS

através de fundos nacionais e quando aplicável co-

financiados pelo FEDER, no acordo PT2020 e projectos

[UID/MULTI/00491/2013]; [UID/MAT/04561/2013];

[PTDC/MAT-NAN/2196/2014].

O problema estudado ocorre numa empresaresponsável pelo estacionamento em Lisboa,quando recolhe os cofres dos parquímetros. Noroteamento para recolha de valores, por se-gurança, as rotas não devem ser antecipáveis.Dada a disposição dos parquímetros é natu-ral considerar um problema de rotas nos ar-cos. Encontram-se estudos sobre este assuntopara roteamento nos nodos, mas não nos ar-cos. Este novo problema de rotas dissemel-hantes nos arcos será designado por DARP (Dis-similar Arc Routing Problem). O DARP é definidonum grafo misto. As arestas representam ruascom dois sentidos servidas numa só passagem;os arcos avenidas que requerem duas passagensem serviço, ou ruas de sentido único. Os no-dos são cruzamentos, becos, ou o depósito, in-ício e fim de todas as rotas. As ligações comcofres para serem recolhidos são designadaspor tarefas. Os serviços são efetuados diari-amente considerando-se um horizonte tempo-ral de cinco dias úteis. Pretende-se determi-nar um conjunto de rotas dissemelhantes, umapor dia, que minimizem o tempo total. Paraobter dissemelhança, as rotas são divididas emperíodos, evitando-se que a mesma tarefa sejaescalonada no mesmo período em rotas difer-entes. Apresentam-se uma formulação em pro-gramação linear inteira mista e uma mataheuris-tica. O resultado de testes computacionais é re-latado.

4. Waste collection planning based on real-time information

Carolina Soares de Morais (carolsoaresdemor-ais@yahoo.com.br) CEG-IST, Instituto Supe-rior Técnico, Universidade de Lisboa, TâniaRamos, Carolina Soares de Morais, Ana PaulaBarbosa-Póvoa

Keywords: Dynamic Routing, Waste Manage-

ment, Sensors, Vehicle Routing Problem with Profits

This paper studies the definition of dynamicroutes on the waste collection problem. Based

on the access to real-time information, providedby sensors located at the waste bin containers,a Vehicle Routing Problem with Profits (VRPP)solution approach is developed considering themaximization of the waste collected while mini-mizing the total distance travelled, resulting in amaximization of profit. Different scenarios arestudied based on real data. The conclusionsclearly show that the usage of real-time informa-tion on containers fill-levels coupled with an op-timization solution approach to define dynamicroutes will potentially increase the profit of thewaste management companies.

5. A biased-randomized heuristic for thehome healthcare routing problem

Maria Isabel Gomes (mirg@fct.unl.pt) Centerfor Mathematics and Applications, Faculty ofSciences and Tecnology, Nova University ofLisbon, Manuel Eliseu, Maria Isabel Gomes,Angel A. Juan

Keywords: biased-randomized heuristic, home

healthcare routing problem, case study

The home healthcare routing problem (HHRP)refers to the problem of allocating and rout-ing caregivers to care-dependent people at theirhomes. It has been mostly tackled in the lit-erature as a rich vehicle routing problem withtime windows. This paper proposes a biased-randomized heuristic, based on the well-knownsavings heuristic, to solve the HHRP. The algo-rithm is tested in small but real-case instanceswhere patients’ visits may occur more than oncea day and, in such cases, all the visits have tobe performed by the same caregiver. The resultsshow the algorithm provides good quality resultsin reasonably low computing times.

Thu.3.1, Thursday, June 29, 15:30-16:50, Room1

Scheduling and Sequencing ISession chair: Rui Borges Lopes

1. Decentralized sequencing in real-time injob scheduling problem

Afshin Mehrsai (amehrsai@inesctec.pt) INESCTEC Porto, Afshin Mehrsai, Gonçalo Figueira,Nicolau Santos, Pedro Amorim, Bernardo

IO 2017 47

SESSIONS

Almada-Lobo

Keywords: Job Scheduling, Decentralized se-

quencing, Industry 4.0

Sequencing of jobs in the queues of machinesis the second task of job scheduling after ma-chine allocation. Conventionally, both alloca-tion and sequencing tasks are statically mod-eled and solved with centralized solutions. To-day, the digital industrial revolution "Industry4.0" facilitates the real-time collection and anal-ysis of data in manufacturing. To tap into this,job scheduling can be explored for decentralizedallocation and sequencing as alternative to cen-tralized solutions. In this study some decentral-ized sequencing algorithms are tested by simu-lating a job shop scheduling case for their perfor-mance and applicability. The results are promis-ing compared to the central solution.

2. Optimizing staff scheduling: a case atINEM

Hendrik Vermuyten (hen-drik.vermuyten@kuleuven.be) CEG-IST,Instituto Superior Técnico, Universidadede Lisboa, KULeuven, Joana NamoradoRosa, Hendrik Vermuyten, Inês Marques, AnaBarbosa-Póvoa

Keywords: emergency medical service, staff

scheduling, optimisation

Staff scheduling problems englobe complex andtime-consuming activities since they rely uponboth work and legal issues and upon individ-ual preferences. Mathematical programmingis a valuable tool to build automatic generatorschedules which can entail economic and so-cial benefits for a company. These problemsare present in many real-world environmentsnamely in the health sector. This research fo-cus on a staff scheduling problem in the emer-gency medical service at INEM in Portugal. Itcorresponds to a 24/7 service of 8 working hoursshift of two services, dispatch center (CODU)and emergency vehicles. We propose a mixedinteger linear programming approach to addressthis staff scheduling problem motivated by spe-cific demand requirements and constraints thatneed to be meet. Models are solved using C++

with the CPLEX solver and this approach is ap-plied to real-life data provided by INEM.

3. Heurística de escalonamento para um lab-oratório de análises de vinhosSara Martins (spsdm@ua.pt) DEGEIT-UA,Sara Martins, Carina Pimentel, MarleneAmorim

Keywords: Serviços, Escalonamento, Heurística,

FJSSP, Tempos de resposta

A globalização tem agitado o panorama indus-trial implicando o reajustamento das organiza-ções, de forma a que estas sejam ágeis e efi-cientes face à volatilidade existente. Contudo,também o setor terciário tem sofrido alteraçõesa este nível, uma vez que assume um papelpreponderante no desenvolvimento da econo-mia. Posto isto, o presente trabalho insere-senum serviço do qual parte da fileira vitiviní-cola nacional vê a sua atividade económica de-pendente. O serviço é caraterizado pela exe-cução de diferentes tipos de análises de vinhos,pretendendo-se, escalonar numa base diária asatividades (ou operações) do laboratório da or-ganização, de forma a minimizar o makespane diminuir os tempos de resposta ao cliente.O problema abordado retrata um caso partic-ular da literatura, Flexible Job-shop Schedul-ing Problem (FJSSP), exibindo uma complexi-dade acrescida devido à diversidade de análises,equipamentos e rotas existentes no shopfloor.Para tal, foi desenvolvida uma heurística paraescalonar a atividade do laboratório, que consid-era as restrições inerentes ao sistema em estudo,procurando, também, incorporar caraterísticasdesse mesmo sistema com vista à obtenção deuma solução de maior qualidade.

4. Heuristic approaches for a flexible job-shop scheduling problemRui Borges Lopes (rui.borges@ua.pt) DEGEIT/ CIDMA, Universidade de Aveiro, SandraSousa, Rui Borges Lopes

Keywords: scheduling; flexible job-shop; re-

sources capacity; heuristics.

This communication addresses a flexible job-shop scheduling problem (FJSP) with a newset of processing characteristics: release dates,

48 IO 2017

SESSIONS

precedence constraints (either between oper-ations of the same job or between differentjobs), resources capacity (including downtimes,changes in capacity, and infinite capacity), andsequence-(in)dependent setup times. The goalis to find a schedule minimising the total num-ber of tardy jobs. To tackle the newly proposedFJSP some heuristic approaches are put forward.The proposed heuristics were evaluated using anewly devised set of instances (some of thembased on real-world data). Main conclusionsand future research opportunities are also pre-sented.

Thu.3.2, Thursday, June 29, 15:30-16:50, Room2

Multicriteria IIISession chair: Alberto Martinho

1. Optimising costs and risks in asset man-agement

Luis Dias (gei12061@fe.up.pt) Faculdade deEngenharia do Porto, Luis Dias, ArmandoLeitão, Luis Guimarães

Keywords: Risk Based Maintenance, Genetic Al-

gorithms, Monte Carlo Simulation, Multi objective,

Health Index, Remaining Useful Life, Multi Criteria

With the increasing age of assets related to op-erational functions, asset management has be-come increasingly more relevant. There is a needto extract the most value from the assets be-fore they are retired from their functions, andso maintenance policies are formulated to pro-long the life of said assets. In this work, we con-sider a generic asset management problem ca-pable of being applied to several kinds of assets.This problem takes into consideration the assetcondition based on a discrete set of states whichevolve with the passage of time modelling assetdegradation. The goal is to determine the op-timal degradation state which preventive main-tenance should be performed. The problem isformulated as a multi-objective search aiming atsimultaneously optimising two objectives of in-terest: risks and costs. For a close approxima-tion to reality, the predictive model describingthe evolution of the degrading system is based

on the use of the Monte Carlo (MC) simulation.Maintenance policies are generated using a Ge-netic Algorithm (GA). The coupled (GA+MC) willbe the key to establish the maintenance policiesthat are able to optimize risks and costs.

2. Investment portfolio formation via multi-criteria decision aid: A brazilian stock mar-ket studyJéssica de Freitas (jgaldinofreitas@gmail.com)Federal Fluminense University, Marcio Basilio,Jéssica de Freitas, Milton Kämpffe, RicardoBordeaux-Rego

Keywords: Stock portfolio; Principal Components

Analysis; PROMETHEE

Acknowledgement: We thank the Federal Flumi-

nense University (UFF) and Coordination of Improve-

ment of Higher Level Personnel (CAPES) for the sup-

port of the research.

The purpose of this paper is to identify how Mul-ticriteria Decision Aid (MCDA) can assist the in-vestment portfolios formation, increasing the re-liability of decision making. In order to developthis paper, a simulation-based approach wasused. Information about the assets traded on thespot market of the São Paulo Stock Exchange -BM&FBOVESPA were selected. They had a 100%of participation in the 246 trading sessions car-ried out in 2015 and had an average number ofbusiness/day greater or equal to 1000. The strat-ification resulted in the selection of 111 assets.Aiming assets evaluation, data were collectedfrom 21 financial indicators. Subsequently, thePrincipal Component Analysis (PCA) was usedto reduce the mass of collected data without theloss of essential information. PROMETHEE IImethod was used for assets ranking; it belongsto the group of methods for MCDA. At the end ofthese stages, four groups of investment portfo-lios were created for simulation. It resulted in acomparison in which it was observed that 100%of portfolios showed positive returns on the in-vestment. The result of portfolios’ group com-posed of assets based on the 21 financial indica-tors were higher than Ibovespa result in the sameperiod.

3. Location-routing problems: Sectorizationand multicriteria

IO 2017 49

SESSIONS

Alberto Martinho (al-berto.m.martinho@inesctec.pt) INESCTEC, Alberto Martinho, Eduardo Alves, AnaMaria Rodrigues, José Soeiro Ferreira

Keywords: Location-Routing Problems, Mul-

ticriteria, Sectorization, Electrostatics, Simulated-

Annealing

Logistic decisions, involving the location of fa-cilities in connection with vehicle routing, ap-pear in many contexts and applications. Givena set of potential distribution centers (DC) anda group of clients, the choice of which DC toopen together with the design of a number ofvehicle routes, satisfying clients’ demand, maydefine Location-Routing Problems (LRP). Thispaper contributes with a new method, the 4-Phase Method (4-PhM) to deal with LRP. Rele-vant advantages of 4-PhM are its generality, thepossibilities of handling multicriteria and of fac-ing large dimension problems. This last apti-tude is a consequence of the sectorization phase,which permits a simplification of the solutionspace. Two procedures are used to construct sec-tors, based on SectorEl, an approach to sector-ization inspired by electrostatics, and they bothuse Simulated Annealing. In the last phase, theresults obtained are evaluated using multicrite-ria analysis. Decision makers play here an im-portant role by reflecting preferences in a pair-wise comparison matrix of the Analytic Hierar-chy Process. Computational results, based onrandomly generated instances, confirm the ex-pectations about 4-PhM and its potentiality todeal with LRP.

Thu.3.3, Thursday, June 29, 15:30-16:50, Room3

Energy, Environment, Natural Re-sources, and Weather IISession chair: Ricardo Lima

1. A framework for suitable prices: the roleof alternative tariff structures in water sup-ply projects

Francisco Silva Pinto (frcsilvap-into@tecnico.ulisboa.pt) CESUR, CERIS,Instituto Superior Técnico, Universidade

de Lisboa, Francisco Silva Pinto, Rui CunhaMarques

Keywords: tariff structures; stochastic desalina-

tion cost; multiobjective optimization; sustainability;

water supply

Water utilities face different challenges that mayforce them to seek diversified prioritized objec-tives. When doing so, particular projects mayhave to be developed, being important to under-stand their impact on water tariffs, and thus, oncustomers. The resulting cost and revenue vari-ability demand a comprehensive study. If thefirst may require a stochastic modeling (in ma-jor cost components) in order to consider its in-herent uncertainty, the second needs to be mod-eled in accordance with context-specific objec-tives set by decision-makers. For that purpose, amulti-objective optimization model (i.e., Frame-work for Suitable Prices) was built to deal withthose diversified requirements (e.g., stochasticenergy costs, affordability, cost recovery, or ad-ministrative simplicity). The model is solvedthrough achievement scalarizing functions withseveral weighting coefficients for a referencepoint, providing a significant perception of thepossible revenue options (and their impact) tothe decision-makers. The proposed method isapplied to a case-study, Boa Vista Island in CaboVerde, in which the background characteristics,namely water sources availability (e.g., the adop-tion of desalination technologies), economic de-velopment and other contextual factors wereconsidered. The key role of tariff structure selec-tion is displayed, instead of assuming it a priori,giving important insights regarding project fea-sibility.

2. Bus fleet management optimization us-ing the augmented weighted Tchebycheffmethod

William Machado Emiliano(williammachado16@gmail.com) Univer-sity of Minho, William Machado Emiliano,Lino Costa, Maria Sameiro Carvalho, JoséTelhada

Keywords: Augmented Weighted Tchebycheff,

50 IO 2017

SESSIONS

Buses Fleet Management Problem, GHG Emissions.

This paper presents a multi-objective optimiza-tion model for the buses fleet managementproblem. This model is solved using the Aug-mented Weighted Tchebycheff method. Theaim is to minimize three objective functions, Z1(CO2 emissions), Z2 (Others type de emissions)and Z3 (total costs), for a bus fleet that usesfour types of buses: diesel, electric bus, electricbus of fast charge, and Compressed Natural Gas(CNG). A public transport company of Joinville,Brazil, where it operates three different PT lines,owns the fleet. The respective data was mod-elled and optimized using the MS Excel solver.Results provided interesting insights concerningthe most adequate strategy for bus selection ac-cording with public transport line characteris-tics and taking into account trade-off betweencosts and emissions. The results indicate thatoptimal solutions include the diesel in the Itingaline and the CNG in the South line. The elec-tric bus is more adequate in the South-North linedue to the large number of stops and low aver-age speed. However, when the costs were disre-garded in some scenarios the best option was theelectric bus for all lines.

3. Risk-averse stochastic programming vs ro-bust optimization and sample average ap-proximation for a virtual power plant

Ricardo Lima (ricardo.lima@kaust.edu.sa)King Abdullah University of Science and Tech-nology - KAUST, Ricardo Lima, Loïc Giraldi,Antonio Conejo, Ibrahin Hoteit, Olivier LeMaître, Omar Knio

Keywords: Stochastic programming, Robust op-

timization, sample average approximation, risk man-

agement, virtual power plant.

In this contribution, we compare risk-aversestochastic programming (SP) with robust op-timization (RO) to address the self-schedulingand market involvement of a virtual power plant(VPP). The decision-making problem involvesuncertainty in the wind speed and electricityprices forecast. The electricity prices forecastsare sampled from an ARIMA model, and thewind speed forecasts are based on a 51-member

wind ensemble. We analyze both approachesconcerning the formulation, uncertainty andrisk, and decomposition methods. The Condi-tional Value at Risk (CVaR) is used as a risk mea-sure in SP because it resembles worst-case mea-sures, which naturally links to RO. We report thecomputational performance of the SP and ROdecomposition methods used. We further com-pare the operational results obtained with SPand RO on the values of the first stage decisionvariables. The objectives of using sampling av-erage approximation are fourfold 1) to provideconfidence intervals for upper and lower boundson the objective function values; 2) to assess thequality of the first-stage solutions; 3) to com-pare first-stage solutions from SP and RO; and 4)to provide an empirical framework to assess theimpact on the results of the number of the con-sidered scenarios.

Thu.3.4, Thursday, June 29, 15:30-16:50, Room4

Metaheuristics ISession chair: Ana Paula Teixeira

1. Solving large-scale railway line planningproblems using a genetic algorithm

Carlos Iglésias (carlos.iglesias@siscog.pt)SISCOG-Sistemas Cognitivos, SA, CarlosIglésias, Ana Carvalho, Susana Brandão,Ricardo Saldanha

Keywords: line planning, lines, frequencies, rail-

way, genetic algorithm

Line Planning (LP) is a fundamental part inthe design of public transportation services, asmany networks are reaching saturation, requir-ing new approaches to reduce operational costsand increase service quality with the availableresources. In the railway context, it requiresthe selection of service lines and frequencies ina network composed of stations linked by net-work segments. These lines and frequencies are,then, used to create a timetable. The goal isto cover a given travel demand which is knownfor origin-destination stations in the so-calledOD matrix. This raises difficulties when esti-mating the projected demand over each line as

IO 2017 51

SESSIONS

it depends on all alternative routes that pas-sengers can take. Although not being linearlyrelated, projected demand, lines and frequen-cies must be estimated jointly. We introduce anew approach to estimate lines’ projected de-mand by introducing a limited pool of preferen-tial routes between origin-destination stationsthat passengers choose based on personal pref-erence or schedule. A fleet size constraint isconsidered, leading to further entanglement be-tween projected demand estimation and line’scapacity. We use a biased random key genetic al-gorithm and introduce specialized crossover andmutation operators for the LP problem. We showcompelling results for a major European Coun-try.

2. An heuristic approach to product linetrimming decisionsXavier Andrade (xavier.andrade@fe.up.pt)Faculdade de Engenharia da Universidadedo Porto, Xavier Andrade, Luís Guimarães,Gonçalo Figueira

Keywords: Meta-heuristics, product-line trim-

ming, operations and marketing integration.

This work exposes an exploratory research intothe product-line trimming decision problem. Byassuming the position of a player in a compet-itive market, we consider the effects of remov-ing products from our mix. Product removal re-sults, usually in a loss of sales which can be com-pensated by reduced inventory and manufactur-ing complexity costs. In order to further studythis problem, we generate world-size instancesand develop a meta-heuristic algorithm, in orderto determine which products to trim, and whichproducts to keep while maximizing profit. Lastly,we discuss the implications of the instances onthe performance of the approach and set aim onfuture research.

3. Ambulance dispatching and relocationproblemAna Sofia Carvalho(asofia.fcarvalho@gmail.com) CMAF-CIOUniversidade de Lisboa, Ana Sofia Carvalho,Maria Eugénia Captivo, Inês Marques

Keywords: emergency medical service, ambu-

lances, location, dispatching, relocation, optimization,

decision-making

Emergency Medical Service (EMS) is one of themost important health care services since it isresponsible for providing basic medical care forany person in an emergency situation. EMShas to manage and mobilize several resources,e.g. high specialized equipment and highlyskilled staff and aims at providing an effectiveand efficient response when an emergency oc-curs. Thus, this is an issue that concerns soci-ety. Three levels of decision can be identified:strategic, tactical and operational. The decision-making process is very important as decisionsmade are usually closely related, having a directimpact in the service’s quality. We focus on theoperational level by solving the dispatching andrelocation ambulance problems. The first de-terministic models proposed in the mid 1970’sbecome more sophisticated through the years.Furthermore, approaches which include exactmethods, heuristic algorithms and simulationhave been developed to include real life featuresnamely the huge level of uncertainty involvedin the emergencies’ occurrence, the populationdensity and the traffic conditions. This workconsiders the Portuguese case of EMS wheresolving these problems has been a handmadetask. We apply optimization approaches to helpmanagers in the decision-making process at theoperational level tasks.

4. Diabetes: Introduction to the analysis ofthe temperature of the foot

Ana Paula Teixeira (ateixeir@utad.pt) Univer-sidade de Trás-os-Montes e Alto Douro, AnaMargarida Matos, Ana Paula Teixeira, Ana Is-abel Pereira

Keywords: Thermography, diabetic foot, temper-

ature, Genetic Algorithm, Simulated Annealing

This work aims to contribute to the preventionand early diagnosis of diabetic foot, in DiabetesMellitus patients. In this study, an analysis of thestate of the art on the use of infrared thermogra-phy to characterize the behavior of the temper-ature distribution in the foot plant, both on nondiabetic and diabetic feet, as well as on the useof optimization techniques to characterize the

52 IO 2017

SESSIONS

feet temperature distribution is presented. Twoglobal non-linear optimization methods, the Ge-netic Algorithm and the Simulated Annealing,were used to characterize the temperature varia-tion of feet using the best known model; thermalplantar images were used. The numerical resultsobtained with both methods for the different re-gions of the feet are presented and a compari-son between those results is made. The GeneticAlgorithm is the one that presented the best re-sults.

Thu.3.5, Thursday, June 29, 15:30-16:50, Room5

Discrete Optimization, Graphs,and Geometry IISession chair: Marta Castilho

1. Models and advanced optimization algo-rithms for the integrated management of lo-gistics operationsTelmo Pinto (telmo@dps.uminho.pt) Universi-dade do Minho, Telmo Pinto, Cláudio Alves,José Valério de Carvalho

Keywords: vehicle routing; loading constraints;

computational study

Acknowledgement: This work has been sup-

ported by FEDER funding through the Programa Op-

eracional Factores de Competitividade-COMPETE and

through the Portuguese Science and Technology Foun-

dation (FCT): project UID/CEC/00319/2013 and grant

SFRH/BD/73584/2010

In this paper, we present the main contribu-tions of the first author’s Ph.D. thesis undersupervision of C. Alves and J. Valério de Car-valho. In that thesis, we propose a set of al-gorithms regarding real combinatorial optimiza-tion problems in the context of transportation ofgoods. These problems consist in the combina-tion of the vehicle routing problem with the two-dimensional bin-packing problem, which is alsoknown as the vehicle routing problem with two-dimensional loading constraints. We also ana-lyzed two related problems, namely the elemen-tary shortest path problem and the vehicle rout-ing problem with mixed linehaul and backhaulcustomers. In both problems, two-dimensional

loading constraints are explicitly considered.Two column generation based approaches areproposed for the vehicle routing problem withtwo-dimensional constraints. The elementaryshortest path problem with two-dimensionalconstraints is addressed due to its importancein solving the subproblem of the column gen-eration algorithms. To the best of our knowl-edge, we contribute with the first approach forthis problem, through a variable neighborhoodsearch algorithm. In what concerns the vehi-cle routing problem with mixed linehaul andbackhaul customers and two-dimensional load-ing constraints, different variable neighborhoodsearch algorithms are proposed.

2. Heuristic algorithms for a capital budget-ing problem with the option to defer

Anabela Costa (anabela.costa@iscte.pt)ISCTE-IUL/ CMAF+CIO, Anabela Costa, JoséPaixão

Keywords: Real options; Capital budgeting;

Scenario-based optimization; 0-1 Integer program-

ming; Surrogate Relaxation; Heuristic

We present and discuss heuristic algorithms forthe capital budgeting problem with the optionto defer the investment decision (CBPD) whereone deals with a set of projects candidates forinvestment over a time period and limited by abudget. The CBPD can be formulated as an in-teger linear programming model. The decisionis taken in order to maximize the total amountexpected and respecting the budget. The valueof each project is estimated by an option pric-ing approach leading to a binomial tree whichgrows exponentially with the number of peri-ods. Since the decision variables are associatedto the leaves of the tree, the integer linear prob-lem turns out to be computationally quite in-tractable even for a small number of projects orperiods. Recently we presented a surrogate con-straint approach providing upper bounds on theoptimal value and sub-solutions that, throughan heuristic procedure are made feasible for theCBPD producing lower bound values for the op-timum. However, our computational experienceshows that the quality of the feasible solutionsis reasonably poor for the largest test instances.

IO 2017 53

SESSIONS

Hence, alternative heuristics should be consid-ered for the problem such as those presented inthis talk. Preliminary computational results arediscussed.

3. Reducing geometric complexity of irregu-lar pieces for nesting problems.Pedro Rocha (pedro.f.rocha@inesctec.pt) IN-ESCTEC, Pedro Rocha, José Oliveira, MariaCarravilla

Keywords: Nesting, framework, geometry, NFP

This work addresses geometrical challenges ofthe Nesting problem, mainly the geometry of the2D pieces and containers involved. In real worldscenarios, some of the pieces have very com-plex geometry, with high number of vertices andedges which creates difficulties for current algo-rithms to process, without reducing their qualityusing approximations. The simplification pro-cess is tackled by this work, which modifies someexisting algorithms to produce simplified highquality representations of its components.

4. New approach for optimization of con-struction and demolition waste manage-ment: Application to the Lisbon metropoli-tan areaMarta Castilho Gomes(marta.gomes@tecnico.ulisboa.pt) CERIS,CESUR, Instituto Superior Técnico, Universi-dade de Lisboa, António Rebello de Andrade,Marta Castilho Gomes, Joaquim Duque

Keywords: Construction and demolition waste,

Lisbon Metropolitan Area, mixed-integer linear opti-

mization, process systems engineering

The growing concern regarding global sustain-ability is increasingly evident in the construc-tion sector due to the impact of waste produc-tion. The aim of this work is to develop a newoptimization approach in order to plan a net-work for recycling Construction and Demoli-tion Waste (CDW). The approach is based onthe methodology of Process Systems Engineer-ing (PSE), and allows better problem systemati-zation and definition of materials flow betweenprocesses of the network. The mixed-integerlinear programming (MILP) model developed isa tool to support CDW management in assess-

ment of the recycling network from a regulatoryperspective (which aims to minimize the totalcost) and from the perspective of transformingentities (aiming to maximize profit of the recy-cling processes). Various scenarios can be de-signed and sensitivity analyses performed, mak-ing the aforementioned assessment clear regard-ing location, types and capacities of the new pro-cesses to be installed. The model was appliedto the 211 parishes that comprise the LisbonMetropolitan Area (LMA), but due to its genericformulation can be applied elsewhere at regionalor national level. The results indicate a prefer-ence for direct deposition in landfills and the factthat high quality recycling processes are not eco-nomically viable for the LMA.

Thu.3.6, Thursday, June 29, 15:30-16:50, Room6

OR TeachingSession chair: José Soeiro Ferreira

1. GLP-Tool: uma ferramenta gráfica,dinâmica e interativa para introdução àprogramação linear

Susana Fernandes (sfer@ualg.pt) FCT - Uni-versidade do Algarve, Susana Fernandes, JoséPereira

Keywords: Programação Linear, Ferramenta

didática Interativa

A GLP-Tool é uma ferramenta didática pen-sada como facilitador do processo de ensino-aprendizagem em Programação Linear (PL),numa abordagem de metodologias ativas. Asrepresentação e resolução gráficas de proble-mas de programação linear a duas variáveispermitem introduzir de forma intuitiva todosos conceitos de PL. A representação gráfica deproblemas de programação linear a duas var-iáveis é de facto muito útil, mas não chega...Por exemplo, sem uma ferramenta dinâmicanão é fácil mostrar/perceber o que se passaquando se alteram coeficientes das restrições ouda função objectivo. A GLP-Tool é uma apli-cação gráfica, interativa e dinâmica onde as in-formações gráfica e numérica são atualizadasem tempo real. Apresentaremos como as fun-

54 IO 2017

SESSIONS

cionalidades da GLP-Tool (que permite explorardiferentes funções objectivo e conjuntos de re-strições, obter informação gráfica e numérica desoluções óptimas e realizar intuitivamente pós-otimização e análise de sensibilidade) facilitam aapropriação de conceitos como: solução admis-sível / região admissível; restrição redundante;solução básica; solução ótima; restrição ativa;soluções múltiplas; região admissível ilimitada;soluções ilimitadas; problema ilimitado; regiãoadmissível vazia / problema impossível.

2. Estudo europeu em investigação opera-cional no ensino superior: resultados impor-tantes e tendências preliminares

João Miranda (jlmiranda@estgp.pt) ESTG, In-stituto Politécnico de Portalegre; CERENA/IST,Ana Paula Teixeira, João Miranda, Pato Mar-garida

Keywords: Educação em IO; Estudo Europeu; re-

sultados; tendências

No âmbito do estudo europeu em Investi-gação Operacional (IO) no Ensino Superior(https://ec.europa.eu/eusurvey/runner/-ORMSeducation ) e cujo relatório foi publicadorecentemente, em dezembro de 2016, são apre-sentados resultados importantes e tendênciaspreliminares. A principal finalidade do estudoé contribuir para uma visão alargada sobre oestado atual da Educação em IO na Europa,existindo um bom apoio das comunidadesnacionais de IO para o questionário da primeirafase. Assim, obteve-se um número significativode respostas: 191, cerca de 30% das quaisidentificaram a instituição ou país da afiliaçãoprincipal. Com base no referido questionário,pretendeu-se conhecer os principais aspetosrelativos a: matrícula, redução da taxa deinsucesso dos estudantes de primeira matrícula,promoção da continuidade, relevância de cursose transição de diplomados para o mercado detrabalho. Atendendo a uma primeira análisedos resultados e tendências, é apresentada umavisão geral da situação atual da Educação emIO nos países da associação EURO. Tambémsão discutidos tópicos importantes e específicospara alguns dos países europeus, bem comose apresentam propostas de desenvolvimentos

futuros.

3. Investigação operacional e artes marciaisJosé Soeiro Ferreira (jsf@inesctec.pt) FEUP -Faculdade de Engenharia, Universidade doPorto e INESC TEC, José Soeiro Ferreira

Keywords: Artes Marciais, Investigação Opera-

cional, Paradigmas Hard e Soft, Intuição, Ética

Artes Marciais (AM) são um sistema global detreino, que vai muito para além do combate.AM são estudadas e praticadas há milhares deanos e a sua influência no mundo é impres-siva. A IO (que tem uma origem marcial) ape-nas agora está a escrever história, tem um im-pacto reconhecido diminuto, quase ninguém aconhece e quem toma decisões marcantes nãoa usa. Muito se escreveu sobre a relevânciadas AM, do ZEN ou da "Arte da Guerra, de SunTzu" para a Gestão, Estratégia ou Liderança. Oobjetivo aqui, é refletir sobre como a sabedo-ria das AM poderá contribuir para os sérios de-safios que a IO enfrenta, face à crescente com-plexidade e volatilidade dos problemas. Tal con-tribuição será visível na articulação de com-petências variadas e complementares. A comu-nicação evidenciará: - consonâncias entre IOe AM; efetivamente ambas envolvem pessoase visam compreender situações problemáticas,holisticamente, e com elas lidar, resolvendo-ase/ou dissolvendo-as; - o ensino e a prática doCorpo e da Técnica em AM e em IO e suas lim-itações; a Mente, como paradigma soft; - o pa-pel da Intuição e das Emoções, incessantementeconfirmado em AM; a Ética presente. Há umcaminho de AM para a IO!

Fri.1.1, Friday, June 30, 9:30-10:50, Room 1

Scheduling and Sequencing IISession chair: Ana I. Pereira

1. Scheduling of nonconforming devices: thecase of a company in the automotive sectorMariana Nogueira (araujomari-ana@hotmail.com) Department of Productionand Systems, University of Minho, Guimarães,Portugal, Mariana Nogueira, Maria SameiroCarvalho, José Oliveira

Keywords: Quality, Complaints, Priorization,

IO 2017 55

SESSIONS

Scheduling, Lekin

This article presents a project developed ina company’s quality department aiming atscheduling nonconforming devices analysis pro-cess. The company faced a problem of no com-pliance with pre-established time requests witha huge impact in terms of fines paid to its de-manding customers of the automotive sector.In order to overcome this problem, schedulingtools were developed and tested. The problemfound in the scheduling is characterized as: theobjective is to minimize the number of tardytasks in identical parallel machines. The simula-tion of different scheduling rules allowed to con-firm that the current prioritization rule is not themost effective one. Preliminary simulation werecarried out using Lekin software showing thatothers criteria promote better results. Accordingto the results achieved from the new developedalgorithm that combines two criteria, it was ob-tained a 7% reduction in the analysis times andconsequently a reduction on the fines paid tothe clients. Despite the preliminary status ofpresent results, it is possible to foresee some im-provements in the analysis process performanceby using decision making support tools basedon scheduling algorithms. This way, a signif-icant improvement on the number of analysiswhich fulfill the defined pre-requirements willbe achieved.

2. Aplicação de meta-heuristicas para mini-mização de lesões músculo esqueléticas rela-cionadas com o trabalho

Arminda Pata (armindapata@ua.pt) Universi-dade de Aveiro, Arminda Pata, Ana Moura

Keywords: Programação de horários, problemas

de afetação e sequenciamento, avaliação dos riscos de

LMERT.

No planeamento e organização do trabalho,um dos maiores problemas consiste em es-tabelecer as afetações mais apropriadas en-tre os recursos humanos e os postos de tra-balho disponíveis. A ideia passa por minimizarhipotéticas lesões músculo esqueléticas de umaequipa de trabalho com tarefas repetitivas emonótonas. Este planeamento pode tornar-se

uma tarefa árdua, dado que não se encontramestabelecidas as causas de aparecimento daslesões músculo-esqueléticas relacionadas com otrabalho (LMERT) assim como os valores lim-ites de exposição. Neste contexto, surge o prob-lema de programação integrada de horários erecursos organizacionais (PPIHRO). Cada postode trabalho pode ter associadas várias causasque originam lesões e cada causa pode ser co-mum a vários postos de trabalho. O ideal é quea exposição dos trabalhadores a estas fontes dedano deva ser distribuída de forma equilibrada.Desta forma, este trabalho propõe uma abor-dagem para a resolução deste problema, combase na aplicação de meta-heurísticas não pop-ulacionais, resultando numa abordagem híbridaque apresenta a melhor solução de afetação deacordo com vários parâmetros definidos por umdecisor. São também apresentados problemasde teste com instâncias reais e respetivos resul-tados, provando-se que na maior parte dos casosse obtém a solução ótima.

3. Simultaneous management of sequencedependent setups and work in progress in aflexible job shop

Nicolau Santos (nicolau.f.santos@inesctec.pt)INESC TEC/ FCUP, Nicolau Santos, GonçaloFigueira, João Basto, Bernardo Almada-Lobo,João Pedro Pedroso

Keywords: Scheduling, Flexible Job Shop, Meta-

heuristics

The flexible job shop has a key role in the mod-elling of modern production systems. In ourstudy we consider a multiple route flexible jobshop scheduling problem, where a given job pro-duction sequence may have alternative routesof variable size. This problem appears in pro-duction environments where machines can beprepared with multiple tools or are configuredby default with different capabilities (e.g. print-ing plants). We also consider costs related tosequence dependent setup, earliness and tardi-ness, a combination seldom seen in the litera-ture. These problems are very hard to solve andtraditionally an iterative allocation/schedulingmethod is used to obtain near optimal solutions.In this work we present a heuristic developed

56 IO 2017

SESSIONS

to handle problems of a real production system,with a very large number of operations. Thealgorithm is characterized by a fast neighbour-hood exploration and the possibility to controlthe work in progress, in order to accumulate sim-ilar operations in the intermediate buffers andreduce the subsequent setup times. The pro-posed method is evaluated in both real instancesand generated problems. A comparison of re-sults with other very well-known heuristics con-firms the quality of the proposed algorithm.

4. Optimal procedures to schedule home carevisits in a health center

Ana I. Pereira (apereira@ipb.pt) InstitutoPolitécnico de Bragança, Marina Matos, FilipeAlves, Ana I. Pereira, Adília Fernandes

Keywords: Genetic Algorithm, Health Care Visits,

Optimization, Scheduling

Currently, it has been verified that populationin Portugal is increasingly aged and it is neces-sary to perform home services. These servicesinclude home visits of patients with impossibil-ity (elderly or sick) of traveling to health cen-ters, where the health professional performs themedical treatments. This service is performedusually by nurses, necessary needing transporta-tion for this purpose. Therefore, it is necessary tomake a schedule of these home care visits, sinceusually this is done manually by the health cen-ter. This work aims to carry out an automaticschedule of home care visits of the Bragan caHealth Center, Portugal, in order to reduce thecosts of travel and the time spent during thetrips. The Genetic Algorithm was used to solvethis problem and obtain different solutions ofhome care visits schedule. In this paper it is pre-sented the schedule of home care visits for threedays, April 18, 19 and 20, 2016 of the Health Cen-ter.

Fri.1.2, Friday, June 30, 9:30-10:50, Room 2

Optimization and Multiple PartiesSession chair: João P. Almeida

1. An analysis of the effects of pricing andgovernment subsidies on patients’ hospitalchoices and social utility through simulation

Aydin Teymourifar (aydinteymuri-far@gmail.com) Anadolu University, AydinTEYMOURIFAR, Onur KAYA, Gurkan OZTURK

Keywords: Health-care system, Simulation,

Emergency service, Public benefit, Patients satisfac-

tion, Optimization

Acknowledgement: This work is supported by

the Scientific and Technological Research Council of

Turkey (TUBITAK) under the grant number 115M185.

In this study, we consider the public and privatehospitals in the health-care system of Turkey. Weanalyze the effects of different pricing strategiesand the effects of government subsidies on thepatients’ preferences regarding hospital choicesand the results of these choices on social utilityand public spending. Public and private hospi-tals in Turkey have different characteristics, suchas different quality and service levels and differ-ent prices. In order to have a balanced the gov-ernment can influence the prices of private hos-pitals through contracts. In these contracts, thegovernment negotiates the prices that will be setby the private hospitals and also agrees to paya certain amount per patient to these hospitals.As a result of this decrease in prices, more pa-tients are directed to private hospitals leading tolower densities in public hospitals and a highersocial utility in general. We analyze these con-tracts and try to obtain the optimal pricing andsubsidy mechanisms considering the effects ofthese decisions on different parts of this system.We develop a simulation mode, considering theemergency services in general, and analyze theeffects of different prices on the patient prefer-ences, waiting times, public spending and socialutility.

2. MCVRP with product-specific time-windows assignmentSara Martins (sara.martins@fe.up.pt) INESCTEC, Faculty of Engineering, University ofPorto, Sara Martins, Manuel Ostermeier,Alexander Hübner, Pedro Amorim

Keywords: Retail, Multi-compartment vehicles,

Vehicle routing, Time-windows, Adaptive Large Neigh-

borhood Search

In grocery retailing, products with distinct tem-

IO 2017 57

SESSIONS

perature requirements need to be distributedfrom the distribution centers to the stores. In thelast years, multi-compartments vehicles, whichmake it possible to physically separate the prod-ucts during transportation, have been deployed.These vehicles can be used by grocery retailersto jointly transport products in the same vehi-cle by using compartments with different tem-perature setups. Nevertheless, these deliver-ies have to respect pre-defined time-windowsat the store level for each product. This workaddresses the multi-compartment vehicle rout-ing problems (MCVRP) with a multi-period set-ting, where a time-window has to be selected foreach store-product pair and used consistentlythroughout the planning horizon. An adaptivelarge neighborhood search (ALNS) is proposedto solve the problem with two groups of opera-tors: daily operators, which focus on the routingimprovement on a daily basis; and weekly oper-ators that looking to the complete planning pe-riod, put the effort in aligning the time-windowsconsistency for each product-store pair.

3. Reference points, which impact can haveupon the reliability of the MCDM methods,TOPSIS methods for Group decision makers& interval data : study and comparisonZhor Chergui (chergui_zhor@hotmail.fr)ENST, Zhor CHERGUI, Moncef ABBAS

Keywords: Group decision makers, Interval data,

Pareto optimal area, Reference points, TOPSIS meth-

ods, Normalization forms

In this paper, two new extensions of TOPSISmethod for Group decision makers and Intervaldata are presented. In particular, the behavior ofsome past contributions when using Nadir pointat the place of Anti ideal point is studied. Other-wise, through statistical studies and simulation,which are mainly based upon smart random in-stances, a comparison between Four algorithmsis carried out, its purpose is to show the most ef-fective one.

4. Hotelling game on networksJoão P. Almeida (jpa@ipb.pt) LIAAD-INESCTEC and Polytechnic Institute of Bragança,João P. Almeida, Alberto A. Pinto

Keywords: Hotelling model, game theory, price

competition, networks, oligopoly models.

We develop a theoretical framework to study thelocation-price competition in a Hotelling-typenetwork game, extending the Hotelling modelwith linear transportation costs from a line (city)to a network (town). We show the existence of apure Nash equilibrium price if, and only if, someexplicit conditions on the production costs andon the network structure hold. Furthermore, weprove that the local optimal localization of thefirms are at the cross-roads of the town.

Fri.1.3, Friday, June 30, 9:30-10:50, Room 3

Optimal Control and Systems Dy-namicsSession chair: Helena Sofia Rodrigues

1. A direct approach to a control problem inpassive tracer advection by viscous point vor-tex flowCarlos Balsa (balsa@ipb.pt) Instituto Politéc-nico de Bragança, Carlos Balsa, Sílvio Gama

Keywords: Keywords:Viscous Point Vortex Flow,

Passive Tracer, Control Problem, Numerical Optimiza-

tion

Vortex dynamics and passive tracers in vortex-dominated flows form a vast area of research thatcontinues to attract the attention of numerousstudies. Among these studies, it has emerged inrecent times a special interest in the use of con-trol theory applied to vortex dynamics. Viscouspoint vortices are singular solutions of the two-dimensional incompressible Euler equations.These solutions correspond to the limiting casewhere the vorticity is completely concentratedon a finite number of spatial points each with aprescribed strength/circulation. By definition, apassive tracer is a point vortex with zero circula-tion. We are concerned with the dynamics of apassive tracer advected by two-dimensional vis-cous point vortex flow. More precisely, we wantto drive a passive particle from an initial start-ing point to a final terminal point, both given apriori, in a given finite time. The flow is orig-inated by the displacement of N viscous pointvortices. More precisely, we look for the optimaltrajectories that minimize the objective function

58 IO 2017

SESSIONS

that correspond to the energy expended in thecontrol of the trajectories. The restrictions areessentially due to the ordinary differential equa-tions that govern the displacement of the passiveparticle around the viscous point vortices.

2. Banking risk as an epidemiological model:an optimal control approachOlena Kostylenko (o.kostylenko@gmail.com)Center for Research and Development inMathematics and Applications (CIDMA),Department of Mathematics, University ofAveiro, 3810-193 Aveiro, Portugal, OlenaKostylenko, Helena Sofia Rodrigues, Delfim F.M. Torres

Keywords: banking risk, contagion spread, epi-

demic approach, optimal control.

The process of contagiousness spread modellingis well-known in epidemiology. However, theapplication to banking market is quite recent.In this work, we present a system of ordinarydifferential equations, simulating data from thelargest European banks. Then, an optimal con-trol problem is formulated in order to study theimpact of a possible measure of the Central Bankin the economy. The proposed approach enablesqualitative specifications of contagion in bank-ing obtainment and an adequate analysis andprognosis within the financial sector develop-ment and macroeconomic as a whole. We showthat our model describes well the reality of thelargest European banks. The main simulationswere done using MATLAB and BOCOP optimalcontrol solver, and the main results are taken forthree distinct scenarios.

3. Agile real-time decision makingRicardo Cunha (gei12049@fe.up.pt) INESCTEC, Ricardo Cunha, Gonçalo Figueira, PedroAmorim

Keywords: Scheduling ; Decentralized Decision

Making ; Real-Time

Customers demand for more variety and aquicker response, which, in addition to the newtechnology that is populating the Industry 4.0,result in increasingly complex and uncertainproduction environments. There is thus theneed for new decision making tools to act on

a real-time basis. Traditional production pro-cess modelling techniques rely heavily on cen-tral decision making structures, which presentnumerous disadvantages when dealing with realsystems, whose agility and responsiveness arefundamental to manage all kind of disturbances.The decentralization of decision making may bean interesting solution to overcome these issues,even though there is a high likelihood that mostof the times the solution found is sub-optimal.This research is focused on the implementationof different optimization methods, both central-ized and decentralized, for a job-shop schedul-ing problem. We are particularly interested inunderstanding the performance of these meth-ods under a dynamic environment, exposed toseveral types of uncertainties. To assess the per-formance of the system, a simulation model thatincorporated the different optimization meth-ods is implemented. Simulations using thismodel enabled to study several solutions explor-ing quick responses to expected events, consid-ering distinct uncertainty factors.

4. Epidemiological models and the diffusionof an advertising campaign

Helena Sofia Rodrigues (sofiaro-drigues@esce.ipvc.pt) Business School,Viana do Castelo Polytechnic Institute andCIDMA - University of Aveiro, Helena Sofia Ro-drigues, Manuel José Fonseca, Jorge EsparteiroGarcia

Keywords: epidemiological models; diffusion

spread, advertising campaign; dynamical system

The understanding of the mechanisms of infor-mation diffusion through the social networks isimportant for marketing purposes; on one hand,we have scarce financial resources; on the other,it is necessary to reach the maximum of individ-uals with the target campaign. In this work, itis studied the use of epidemiological models tounderstand a case study of a viral marketing ad-vertising. It is incorporated a variable spreadingrate and is analyzed the possible effects in thedevelopment of the diffusion of the advertisingcampaign.

IO 2017 59

SESSIONS

Fri.1.4, Friday, June 30, 9:30-10:50, Room 4

Metaheuristics IISession chair: Cláudio Alves

1. Balancing mixed-model assembly sys-tems with a variable neighbourhood descentmethod

Parisa Sadeghi (p.sadeghi@fe.up.pt) Facultyof Engineering, University of Porto, ParisaSadeghi, Rui Diogo Rebelo, José Soeiro Ferreira

Keywords: Mixed-model assembly line balancing,

Footwear industry, Optimisation, Variable neighbour-

hood descent method

This work considers a new Mixed-model Assem-bly System Balancing Problem in a footwear in-dustry in Portugal. Footwear industry in thiscountry has improved drastically, to become oneof the main world players. The flexible trans-portation system of the stitching lines simulta-neously moves various boxes, with pairs of dif-ferent models of shoes, in any direction andto any workstation. The objectives are find-ing a feasible solution, minimizing the num-ber of workstations and smoothing operators’workload. An optimisation model was de-veloped, useful for some purposes, but withlimitations to solve real dimension problems.Therefore an approximate approach by usingan adapted Ranked Positional Weighted (RPW)method which integrated by another generatedheuristic is used to balance the mixed-modelsystem. Then a Variable Neighbourhood De-scent (VND) method is applied as an improve-ment technique. Computational tests were un-dertaken, also based on real data. Simulationwas also used to evaluate the solutions. Theresults obtained with this solution method arequite favourable, when compared with somebalances executed by the company. Moreover,the applied method is easy to use, faster and veryconvenient to analyse the consequences of anyline change and it could be useful for similar in-dustries.

2. Abordagem heurística na avaliação nosumódulo de risco de incêndio em SolvênciaII

Nelson Chibeles-Martins (npm@fct.unl.pt)

CMA / DM - FCT NOVA, Joana Fradinho,Lourdes B Afonso, Nelson Chibeles-Martins,Isabel Gomes

Keywords: Programação Linear Binária, Meta-

Heurísticas, Solvência II, Submódulo de Risco de In-

cêndio

Acknowledgement: This work was partially sup-

ported by the Fundação para a Ciência e a Tecnologia

(Portuguese Foundation for Science and Technology)

through the project UID/MAT/00297/2013 (Centro de

Matemática e Aplicações)

Solvência II revê o regime prudencial para ativi-dades seguradora e resseguradora na União Eu-ropeia com o objetivo de reduzir o risco deinsolvência. O Regulamento Delegado (UE)2015/35 da Comissão de 10 de outubro de 2014contém regras para implementação da Solvên-cia II. De forma a obter o requisito de capitalreferido no art. 132º do submódulo de riscode incêndio é necessário avaliar a maior con-centração de risco de incêndio de uma empresade seguros ou de resseguros correspondente aoconjunto de edifícios com a maior soma se-gurada de todos os edifícios que estão parcialou totalmente situados num raio de 200 met-ros. Consequentemente, ferramentas de Inves-tigação Operacional tornam-se cada vez maisrelevantes na Análise Actuarial. Considere-se oseguinte problema: dada uma lista de segura-dos de um território é necessário encontrar ocentro de um círculo com raio de 200 metrosque agregue a maior soma de capital seguradoconsiderando-se todos os edifícios localizadosdentro desse círculo. Este problema pode serformulado como um Problema de PL Binária,o que acarreta as usuais questões de complexi-dade computacional. Neste trabalho os autorespropõem uma formulação para o referido prob-lema e uma abordagem heurística para a sua res-olução.

3. Variable neighborhood search for inte-grated planning and scheduling problems

Cláudio Alves (claudio@dps.uminho.pt) Uni-versidade do Minho, Mário Leite, CláudioAlves, Telmo Pinto

Keywords: Planning, scheduling, integrated opti-

60 IO 2017

SESSIONS

mization, metaheuristics

In this talk, we address the integrated planningand scheduling problem on parallel and identi-cal machines. The problem combines two othercombinatorial optimization problems which aresolved simultaneously in an integrated form.The first problem consists in determining theset of jobs that should be processed in eachperiod of time. The second problem consistsin assigning the jobs to the machines accord-ing to their release and due dates. We describenew optimization approaches based on localsearch and variable neighborhood search basedon two neighborhood structures. Two differ-ent algorithms were implemented for the con-struction of the initial solutions combining dif-ferent strategies for the definition of the initialsequence of jobs. Computational experimentsare reported using benchmark instances fromthe literature.

Fri.1.5, Friday, June 30, 9:30-10:50, Room 5

Simulation and Stochastic Pro-grammingSession chair: Ricardo Lucas

1. The level of detail in simulation models:why does it matter?Luís Pereira (luis.f.pereira@inesctec.pt) INESCTEC, Luís Pereira, Samuel Moniz, Jorge Pinhode Sousa, Américo Azevedo

Keywords: Systems Design, Simulation Mod-

elling, Process Design, Production Planning, Discrete

Event Simulation (DES)

Acknowledgement: The authors gratefully ac-

knowledge TEC4Growth - Pervasive Intelligence, En-

hancers and Proofs of Concept with Industrial Impact /

NORTE-01-0145-FEDER-000020 financed by the North

Portugal Regional Operational Programme (NORTE

2020), under the PORTUGAL 2020 Partnership Agree-

ment, and through the European Regional Develop-

ment Fund (ERDF).

When developing simulation projects, mod-elling the problem at hand and deciding the re-quired level of detail is no easy task. Many dif-ferent aspects have to be considered, and there

is no standard procedure or established refer-ence guide to follow. The level of detail is notonly critical to the computational performanceof the model and to the time required to developit, but it also severely influences the accuracyof the measured indicators. The confidence onthe model and its efficiency are dependent onthe adopted level of detail. Traditionally, sim-ulation practitioners decide on the level of de-tail based on experience. However, how doesa poor decision in this aspect affect the simu-lation results? Or, more specifically, how canwe relate real world problems with different lev-els of detail? We have focused on typical deci-sions for manufacturing systems design and op-eration, to study the impact of different levels ofdetail when solving those problems. From pre-liminary results, we have defined the adequatelevel of detail for specific decisions in simulationmodels. Researchers and practitioners will ben-efit from such reference as it provides a startingpoint in selecting the level of detail, bypassingthe trial and error phase of simulation projects.

2. A simulation and optimization frameworkfor integrated planning and process designRuben Dias (ruben.r.dias@inesctec.pt) INESCTEC, Ruben Dias, Luís Pereira, Samuel Moniz,Alexandra Marques, Américo Azevedo, JorgePinho de Sousa

Keywords: Industry 4.0, Bottleneck Detection,

Process Design, Production Planning, Discrete Event

Simulation (DES), Simulation-Optimization

Acknowledgement: The authors gratefully ac-

knowledge the support of ERDF - European Regional

Development Fund through the Operational Pro-

gramme for Competitiveness and Internationalisation

- COMPETE 2020 Programme, and of the Portuguese

funding agency, FCT - Fundação para a Ciência e a Tec-

nologia, through project SAICTPAC/0034/2015- POCI-

01-0145-FEDER-016418

The high complexity of today’s manufacturingplants makes the task of predicting the impactof improvement decisions very difficult. Lay-out changes, the production mix, complex andalternative routings, among other aspects, re-quire comprehensive modelling approaches, al-lied with adequate process monitoring and in-

IO 2017 61

SESSIONS

novative optimization strategies, for supportingreliable decision-making processes. Followingthe Industry 4.0 paradigm, this article proposesa robust and practical framework to holisticallyassess decisions concerning process design andproduction planning, from both offline and real-time perspectives. A Discrete Event Simulation(DES) model was built to represent the mate-rial and information flows of a real-world pro-duction system. The system dynamics are mon-itored and controlled through a complete setof KPIs, comprehensive visualization tools, andscheduling rules, used to identify bottlenecksand to balance the production load in real-time.A novel concept of "looking ahead" during sim-ulation was also introduced, as a way to antic-ipate and find the best-fitted decisions for spe-cific scheduling problems occurring during thesimulation. The result is an integrated frame-work, effective in improving overall performancemetrics related to process design and produc-tion planning. The proposed approach provedto be useful for decision-makers to improve theoperational performance of complex productionsystems.

3. A stochastic agent-based model with op-timization strategies for nursing workforceplanning

Mário Amorim Lopes (mario.lopes@fe.up.pt)INESC-TEC, FEUP, Mário Amorim Lopes, Ál-varo Almeida, Bernardo Almada-Lobo

Keywords: Healthcare workforce planning,

Health policy, Agent-based simulation, Optimization

Human resources for health (HHR) are criticalfor delivering health care services. Since thehealth labor market faces many legal and regu-latory rigidities, timely and adequate planning ofHHR is necessary to ensure there will be enoughpractitioners to attend to the needs of the pop-ulation. Nurses, in particular, are a corner-stone of health care services, with a shortage po-tentially leading to unmet care needs. To thispurpose, system-level approaches, such as Sys-tem Dynamics, have been used with reasonablesuccess to generate forecasts and assess futureneeds. However, micro-level approaches allowfor studying behavior at the level of an individual

nurse, opening a whole new realm of researchquestions that may now be addressed. In thiswork we present a stochastic agent-based sim-ulation model to forecast the Portuguese nurs-ing workforce until 2050. Additionally, we useuniversity vacancies as a policy lever for eitherincreasing or decreasing the workforce size, anduse the Monte Carlo method to experiment withobjective functions capable of optimizing theworkforce causing minimal disruption to univer-sities.

4. The PackNStore case studyRicardo Lucas (Ricardo.G.lucas@inesctec.pt)INESCTEC PORTO, Ricardo Lucas, LuísPereira, Samuel Moniz, Américo Azevedo

Keywords: discrete event simulation

Acknowledgement: The authors gratefully

aknowledge: ERDF - European Regional Develop-

ment Fund through the Operational Programme

for Competitiveness and Internationalisation -

COMPETE 2020 Programme, and by National Funds

through the Portuguese funding agency, FCT -

Fundação para a Ciência e a Tecnologia, within

Project SAICTPAC/0034/2015- POCI-01- 0145-FEDER-

016418.

A year without improving is a year won by com-petitors - the famous sentence once said by J.M. Duran is no less true today. Continuous im-provement is a reality in most places, and apartfrom "What can we improve now?", companiestry to answer "What will we need to improvein the future?". Simulation models are power-ful in many ways, and they can predict the im-pact of decisions based on "what if?" scenariosthat the company might face in the near future.The "PackNStore" case study is an example ofan expansion in a packaging and storage facil-ity. Through a discrete event simulation model- which more easily allows the use of complexdecision logic, when compared to other types ofmodelling -, it is possible to recreate the facil-ity and change its initial conditions, namely thenumber of AGV’s and buffer sizes, due to its im-portance for the project. Thus, insights can beobtained regarding the output of the plant, andconclude on how its productivity and efficiencyare affected, supporting decision making. This

62 IO 2017

SESSIONS

kind of analysis using discrete event simulationcan prove itself very useful, even when dealingwith large models, as the one studied.

Fri.1.6, Friday, June 30, 9:30-10:50, Room 6

Augusto Queiróz Novais - ProcessSystems EngineeringSession chair: Tânia Pinto-Varela

1. A decision-support framework for theplanning optimisation of biopharmaceuticalindustrial processes

Miguel Vieira(migueljvieira@tecnico.ulisboa.pt) CEG-IST, Instituto Superior Técnico, Universidadede Lisboa, Miguel Vieira, Pinto-Varela Tânia,Barbosa-Póvoa Ana Paula

Keywords: decision-support, planning optimisa-

tion, biopharmaceutical production

The complexity in enterprise management op-timisation of current industrial processes playssignificant challenges to the implementationof efficient decision-support tools. Followingthe advances in computer technology, manual-based planning/scheduling systems have beenreplaced by integrated computational manage-ment tools for core business processes, often inreal-time, to enable a transversal platform forthe flow of information. Therefore, the imple-mentation of process optimisation support be-comes an important tool to improve produc-tion competiveness and responsiveness to mar-ket variability while addressing multiple prof-itability criteria. Despite the major research ad-vances in the modelling applications to addressthe industrial planning complexity, the abilityto integrate optimisation methods in decision-making frameworks still faces significant chal-lenges. An effective decision-support tool needsto consider, for example, the process network in-terconnectivity, the dynamic operational uncer-tainty and the computational performance hin-drance, allowing the decision-maker to evalu-ate the quality of solutions according to com-pany’s objectives. In this work, we discuss theintegration of a model-based approach for theoptimisation of biopharmaceutical production

planning into a decision-support managementtool and its framework application for a success-ful implementation in real industrial environ-ments.

2. A simulation-optimization framework toaddress the stochastic nature of process de-sign and planning decisions in the chemical-pharmaceutical industryCatarina Marques (eq98022@fe.up.pt) Facul-dade de Engenharia da Universidade do Porto/ INESC TEC, Catarina Marques, Samuel Mo-niz, Jorge Pinho de Sousa, Ana Paula Barbosa-Póvoa

Keywords: Process design; Capacity planning;

Scale-ups; Mixed Integer Linear Programing; Monte

Carlo simulation; Uncertainty modelling; Machine

Learning.

Acknowledgement: The first author gratefully

acknowledges the financial support of Fundação

para a Ciência e Tecnologia (FCT), under the grant

PD/BD/105987/2014. This research was also sup-

ported by TEC4Growth - Pervasive Intelligence, En-

hancers and Proofs of Concept with Industrial Impact /

NORTE-01-0145-FEDER-000020 financed by the North

Portugal Regional Operational Programme (NORTE

2020), under the PORTUGAL 2020 Partnership Agree-

ment, and through the European Regional Develop-

ment Fund (ERDF).

The pharmaceutical industry has been increas-ingly facing pressures from market and regula-tors to be more responsive and efficient. Thehigh investments with low success rates and longdevelopment cycles, together with the signifi-cant stochastic nature of the new product devel-opment (NPD) process, add considerable com-plexity to the product launch planning prob-lem. Both market and technical uncertain-ties arise during this process, with a consider-able impact on the company’s decision-makingprocess. The purpose of this work is to ad-dress this planning problem, considering bothtypes of uncertainty, and integrating process de-sign and capacity decisions, under limited re-sources associated to the need of processingin the same plant products under developmentand products in commercialization. A MixedInteger Linear Programing (MILP) model, com-

IO 2017 63

SESSIONS

bined with a two-step Monte Carlo Simulation(MCS) framework, is presented. The MCS com-ponent captures the simultaneous impact of un-certainty in product demand, and in the out-comes of clinical trials, through normal distribu-tions, and Bernoulli distributions, respectively.Process designs and scale-ups were determinedand machine learning techniques were incor-porated into the model, as a way to reveal andexploit correlations between the different pro-cess design configurations, thus contributing toa more reliable and robust decision-making pro-cess.

3. Continuous-time formulations for multi-product pipeline schedulingPedro Castro (pmcastro@fc.ul.pt) CMAFCIO,Faculdade de Ciências, Universidade de Lis-boa, Pedro Castro, Hossein Mostafaei

Keywords: Mathematical modelling; Optimiza-

tion; Logic; Computational Logistics

Acknowledgement: Financial support from Fun-

dação para a Ciência e Tecnologia through projects

IF/00781/2013 and UID/MAT/04561/2013

Pipelines are the most reliable and cost-effectiveway of transporting large volumes over long dis-tances. They are used in the petroleum sup-ply chain for transporting crude oil from wellsto refineries and for transporting refined prod-ucts to local distribution centers. Mathemati-cal formulations for pipeline scheduling can bedivided into batch and product centric. Batchcentric approaches require the modeler to con-vert the products initially inside the pipelineinto a sequence of old batches, and to spec-ify the number of new batches to be injectedat the system input nodes. On the other hand,product centric approaches do not allow dif-ferent batches of a product in a segment butcan rigorously enforce forbidden product se-quences. In this work, we address the optimalscheduling of straight pipelines with multipleintermediate dual purpose nodes and comparetwo recent continuous-time MILP formulationsderived from Generalized Disjunctive Program-ming. The computational results show that thebatch centric formulation requires fewer timeslots to represent a schedule. Because of this, a

better solution was obtained for Ex1. The prod-uct centric formulation benefits from not requir-ing initial decisions that may compromise solu-tion quality. It improved the makespan by 8%for Ex3. Overall, there is no overall best formu-lation.

4. Integrating stochastic programming withscenario reduction methods to support thedesign of biomass to bioenergy supply chainsunder uncertainty

Helena Paulo (hpaulo@deq.isel.ipl.pt) ISEL,IPL, Rua Conselheiro Emídio Navarro, 1959-007 Lisboa, Helena Paulo, Teresa Cardoso-Grilo, Susana Relvas, Ana Paula Barbosa-Póvoa

Keywords: Supply Chain Design; MILP model;

Bioenergy; Stochastic Programming; Scenario Reduc-

tion Methods

Planning problems involving optimization un-der uncertainty are commonly tackled throughstochastic programming models, with scenariotrees being typically used to model uncertain pa-rameters. Real world problems require a largenumber of scenarios to accurately representtheir uncertainty context which likely results ina computationally intractable problem. The de-sign and planning of biomass to energy supplychains, where more than one source of uncer-tainty need to be addressed simultaneously, is anexample. To overcome these difficulties, an inte-grated approach combining scenario reductionmethods with a stochastic mixed integer linearprogramming model is proposed. Uncertaintyon two main critical parameters are considered:the availability of biomass and the technolog-ical process conversion factors. Two scenarioreduction methods are explored: the reductionmethod proposed by Heitsch and Römisch in2003; and the reduction method proposed byKaruppiah, Martin and Grossmann in 2010. ThePortuguese biomass to energy and chemicalssupply chain is considered as case study. Theresults have shown that: i) uncertainty influ-ences supply decisions, both in terms of loca-tions, capacities, technologies and costs; ii) re-duction methods have potential to substantiallyreduce computational effort, while simultane-

64 IO 2017

SESSIONS

ously yielding a solution value close to the solu-tion obtained with the original full scenario tree.

IO 2017 65

Operação Nariz Vermelho

A Operação Nariz Vermelho é uma Instituição Particular de Solidariedade Social que desde 2002leva alegria à criança hospitalizada, aos seus familiares e profissionais de saúde, através da arte eimagem do Doutor Palhaço promovendo, todas as semanas, visitas às enfermarias pediátricas devários hospitais do país. Este serviço é oferecido aos hospitais e financiado através de donativos,particulares empresariais, campanhas e merchandising.

Os Doutores Palhaços da Operação Nariz Vermelho são profissionais vindos de diferentes áreasartísticas com grande capacidade de brincar e improvisar. No entanto, um hospital não é um circo.Atuar numa enfermaria pediátrica exige competência, profissionalismo e muita sensibilidade. Éda responsabilidade da associação treinar e manter a alta qualidade dos artistas. Hoje a OperaçãoNariz Vermelho conta com uma equipa de 23 profissionais que, além de toda a formação específica,têm ainda formação contínua de procedimentos e rotinas relacionadas com o hospital.

A Operação Nariz Vermelho visita mais de 40.000 crianças por ano em 14 hospitais em Lisboa,Porto, Coimbra e Braga.

Mas o número que mais conta é o número 1..

1 OLHAR DE CADA VEZ.

1 CORAÇÃO DE CADA VEZ.

1 SORRISO DE CADA VEZ.

Visite-nos em http://www.narizvermelho.pt

IO 2017 67

Tourism

ValencaValença is a town from Alto Minho that belongs to the district of Viana do Castelo. The historicdevelopment of Valença is deeply connected with its geographic situation, a strategic location at theborder between the north of Portugal and Spain.The Fortress of Valença - the most emblematic of its icons - marks its presence on the top of a hill,showing the greatness of one of the oldest Portuguese settlements, whose magnificence is mayberesponsible for the first impression we get when we arrive.Valença grew way beyond its walls, but its main square keeps its liveliness and is today almost exclu-sively dedicated to tourism and to a diverse trade, with hundreds of stores, bars, cafés and restau-rants.There are many interesting historical points that worth a visit, inside and outside the fortress. Thetour to the Valença Fortress, especially organized for the conference participants, shows a little ofits history, but if you are interested in visiting other monuments you should address to the tourismoffice inside the fortress, Paiol de Marte, Coroada.

The Fortress of ValençaThe fortress of Valença embraces the Medieval Cathedral of Tui, while the river Minho witnesses theadventures of the borderlands.The fortress, with 5 km perimeter, had its first walls built in the early 13th century, erected on apromontory previously occupied by a celtic fort and a Roman village.The fortress, completed in the 17th/18th century, is today one of the exponents of the Vauban modeland awaits the classification as Cultural Heritage of Interest to Humanity. This is one of the mostextraordinary monuments of European military engineering of the modern age. Being a nationalmonument, it offers one of the largest and best preserved historical quarters of its kind in Europe,with 12 bastions, 5 ravelins, 32 bartizans and 194 gunports as outstanding features.Inside it, we must highlight:

• Church of St Mary of the Angels: Romanesque temple, 13th century;

• Collegiate Church of St Stephen: 13th century, romanesque church which was rebuilt in neo-classical style and was the base for the former Collegiate Church of Valença and the Bishopricof Ceuta. In its interior we highlight the episcopal chair of gothic-mudejar style dated in the15th century;

• Military Chapel of Good Jesus: 17th century, baroque and neoclassical architecture;

• Chapel of Our Lord of the Encounter: 18th century, baroque architecture;

IO 2017 69

TOURISM

• Chapel of Mercy: 16th century, baroque and neoclassical architecture. in its interior we findthe ”Dead Lord” sculpture, created by Master Teixeira Lopes;

• Powder keg of Açougue: rectangular military warehouse surrounded by a high wall with a gatebetween pilasters;

• Powder keg of Mars: 18th century, military warehouse with two rectangular rooms;

• Gates of the Coroada: 17th century, ordered by King D. Pedro II, they are the main entranceto the fortification;

• Roman Milestone: iip, from 1st century A.C., it marks 42 miles distance from Braga to Tui andwas ordered by Emperor Claudius; it also served as the whipping post;

• Courtains of St Francisco: armory of the fortress, to the East, next to the bastions of St Fran-cisco and Faro;

• Governor’s Military Palace: it also held the royal class of artillery of Valença;

• Former Domus Municipalis: 14th century, medieval, with outstanding vaults, made with ter-racotta bricks, and arcades;

• Fountain of the Village: Roqueira fountain from 15th century, medieval, on the West side ofthe fortress;

• Statue of St Theutonius: at the Coroada, opposite the Military Chapel of Good Jesus, it hon-ours the first Portuguese saint born in Valença.

More informationFurther information may be found at the municipality official tourism website:

http://www.visitvalenca.com/

70 IO 2017

EstudIO

Planning home healthcare services - a routing andscheduling problem

Ana Raquel de Aguiar1 Tânia R. P. Ramos1 M. Isabel Gomes2

1Instituto Superior Técnico, Universidade de Lisboa; Email: a.raquel.aguiar@ist.utl.pt2Centro de Matemática e Aplicações, Universidade Nova de Lisboa

Problem StatementThe ageing population alongside a shortage of informal care availability are two of the factors thathave led to an increased need for assisted living support nowadays. Both public-based and privatehome healthcare service providers struggle with both lowering public expenditures and trying tostay competitive in the market, respectively. Moreover, these institutions face daily a scheduling androuting problem - allocating caregivers to each patient, who have several requirements and prefer-ences of their own, and take into consideration other constraints such as continuity of care or work-load balance. These services are frequently time-dependent, adding complexity to the problem.Typically, all the planning is done manually. This work is based on a real case study of a PortugueseHome Social Care Service and aims to plan efficiently the daily work schedule for each caregiver.With a group of 36 patients with different levels of care needs (classified into two typologies: bedrid-den or semi-dependent), the objective is to define the schedule for the 7 days of the week for the twotypes of teams (comprising of either one or two caregivers), meeting all patients requests (such asdaily living activities and/or transportation to the day care center). The number of caregiver visits toeach patient varies from four times a day to once a week.

Problem RelevanceIn a context where the spending with home healthcare is expected to increase, actions must be takenregarding the efficient distribution of scarce resources (such as the caregivers). This problem is ofthe utmost importance for governments, health institutions and solidarity institutions, who needto plan budgets and allocate resources. By optimizing caregivers’ routes while satisfying the insti-tution’s operational specifications, it is possible to increase the number of patients to whom careis provided, while assuring an adequate level of workload assigned to each caregiver. For instance,the reduction of caregivers’ travelling time between patients’ visits may allow the admission of morepatients into the institutions, enhancing the amount care delivery to the community.

IO 2017 71

ESTUDIO

Methodology used to solve the problemThe problem is modelled as a Multi-Period Vehicle Routing Problem with Time Windows with addi-tional constraints. We propose a MILP that addresses real world constraints such as lunch breaks,patient visit frequency, and loyalty between caregiver and patient within a day and within a week.Two objective functions are considered separately: 1) minimizing travelling time, and 2) workloadbalance among teams. This model has been developed in GAMS and solved with CPLEX. Being a NP-hard problem, the large computational times led us to the development of a solution methodology,which consists on running the model independently for the two typologies of patients (bedridden orsemi-dependent) and then, within each typology, run the model once more, independently, for eachday of the week. Firstly, the day with the largest number of patients was solved, and the patient/teamallocation for the remaining days was set according to this first solution.

ResultsThe solution methodology was applied to the real-case study with 10 bedridden and 26 semi-dependent patients. The current solution presents a travelling time of 854 minutes. Notice thatcurrently it is not assured that bedridden patients are visited by teams of two caregivers and theloyalty between caregivers and patients is also not present.

Our solution, when imposing only teams of two caregivers visit bedridden patients and the pa-tient/team weekly loyalty, the travelling time becomes 11% smaller than the currently observed (759minutes). One should emphasise the space there is for improving the efficiency of this service, sincethe 11% reduction is achieve with a much more constrained environment.

In terms of workload, respecting the previously mentioned loyalty, the service time differenceamongst teams of the same type and within a day, has a maximum value of 188 minutes, while theminimum is less than one minute. This presents a considerable reduction when compared with thesolution from the first objective function (maximum value was of 216 minutes and a minimum of 74minutes).

72 IO 2017

ESTUDIO

Impressão tridimensional bimaterial por processo aditivo1

Daniel Bandeira1 Marta Pascoal1,2

1Departamento de Matemática, Universidade de Coimbra, Apartado 3008, EC Santa Cruz, 3001-501Coimbra; Email: danielf.a.bandeira@gmail.com

2Instituto de Engenharia de Sistemas e Computadores de Coimbra; Email: marta@mat.uc.pt

Descrição do problemaA impressão tridimensional é um processo em que um objeto é criado pela adição de sucessivascamadas de material. No processo de estereolitografia a adição das camadas é conseguida a partirde um polímero líquido, que solidifica quando exposto a uma luz laser. Este método tem crescidoem popularidade quando é utilizado um só material. No presente trabalho estuda-se a sua ex-tensão à impressão de objetos em que o polímero solidifica em torno de uma grelha de metal. Aestrutura metálica provoca dificuldades, uma vez que pode gerar zonas “sombreadas” na super-fície polimérica. Estuda-se a distribuição de sistemas de espelhos pelas paredes da impressora,por forma a refletir a luz proveniente do laser com orientações bem escolhidas e, assim, atingir assecções da camada a solidificar. Estes sistemas de espelhos terão as suas posições fixas ao longo daimpressão, por forma a manter a estabilidade do equipamento o mais possível. No entanto, podemser orientados de modo a refletir a luz laser nas direções pretendidas.

Relevância do problemaÉ crescente a procura de componentes complexos que combinem 2 ou mais materiais, e possuamcaracterísticas acrescidas, em termos físicos e de funcionalidade. A técnica desenvolvida pretenderesponder à procura de métodos fiáveis e rentáveis para produzir peças complexas em pequenasérie. Exemplos da sua aplicação são ortóteses personalizadas, componentes inteligentes, ou peçascomplexas ou frágeis, em que a opção sobreinjeção não é viável. Apesar do seu potencial, não seconhecem ainda impressoras capazes de trabalhar a 3 dimensões, empregando mais do que ummaterial.

Metodologia utilizada na resoluçãoO trabalho divide-se em 2 partes: 1. calcular o menor número de sistemas de espelhos, e a suadisposição, por forma a permitir uma impressão completa (PNE); 2. afetar as posições a imprimiraos sistemas de espelhos, em cada camada, obtendo os ângulos de incidência necessários para essaimpressão (PAE). O PNE é modelado como um problema de cobertura. O PAE é modelado como umprograma linear com restrições de afetação e de cobertura, e com 2 critérios: minimizar o númerode sistemas de espelhos acionados por camada e maximizar os ângulos de incidência, procurandominimizar a distorção da luz. Para obter algumas solução deste problema bicritério, utilizam-sesomas pesadas das funções objetivo, variando os pesos. Em alternativa o PAE é formulado como umproblema de cobertura, seguido de uma seleção greedy do ângulo de incidência em cada posição aimprimir (A6).

ResultadosA impressão de um cubo com malha metálica variável foi o caso de estudo para os modelos acima,tendo sido utilizados os solvers Matlab e CPLEX.

1Financiado parcialmente pelo projeto PT2020-POCI-SII & DT 17963: NEXT.Parts, Next-Generation of Ad-vanced Hybrid Parts, através do COMPETE 2020, e pela FCT, através do projeto UID/MULTI/00308/2013.

IO 2017 73

ESTUDIO

O PNE foi resolvido em tempos entre 5.03 e 1323.36 segundos, com 2 a 4 espelhos como soluçõesótimas. O PAE foi resolvido com pesos (1−λ,λ), λ= 0,0.25, . . . ,1, e com A6. A Figura 1 mostra os re-sultados mais relevantes. A6 foi sempre a estratégia mais rápida. Em termos da solução encontrada,λ= 0.5 parece fornecer um bom compromisso entre os 2 critérios.

•×∗¦

4

ä •×∗¦

4

ä •×∗¦

4

ä•×∗¦4ä •×∗¦

4ä

•×∗¦

4

ä2 3 5 6 8 9

2

3

4

Teste

µ1•×∗¦4ä

•

×∗¦4

ä

•×∗¦4ä

•×∗¦4ä•×∗¦4ä •×∗¦4ä

2 3 5 6 8 945

56

67

78

Teste

µ2

(o)

•×∗¦4ä

•

×∗¦4ä •×∗¦4ä •×∗¦4ä

•

×∗¦4ä •×∗¦4ä2 3 5 6 8 9

0

700

1400

2100

Teste

Tem

po

(s)

• λ= 0 × λ= 0.25 ∗ λ= 0.5 ¦ λ= 0.75 4 λ= 1 ä A6

Figure 1: Resultados médios para PAE (µ1: número de espelhos; µ2: ângulo de incidência)

74 IO 2017

ESTUDIO

Sistema de Apoio à Decisão para elaboração de mapas deexames no Ensino Superior

Raphaël Martins Cordeiro1 Maria Antónia Carravilla2 José FernandoOliveira2

1Faculdade de Engenharia, Universidade do Porto, Portugal; Email: ee12081@fe.up.pt2INESC TEC e Faculdade de Engenharia, Universidade do Porto, Portugal; Email:

{mac,jfo}@fe.up.pt

ResumoO problema de calendarização de exames é um problema típico de otimização combinatória quetem a particularidade de ter um número muito elevado de restrições, tornando difícil encontrarsoluções admissíveis para o problema [1]. O objetivo deste trabalho passa pela geração de um cal-endário de exames aplicando a meta heurística BRKGA- Biased Random-Key Genetic Algorithm. Porcalendário entende-se uma alocação dos exames a salas e no tempo. De modo a tornar possível ageração de soluções admissíveis as restrições são divididas em restrições fortes e restrições fracas,as últimas podendo ser violadas, com uma penalização na função objetivo, o que permite encontrarsoluções para o problema.

Descrição do problemaNo problema de calendarização de exames abordado neste trabalho, pretende-se atribuir os examesde um curso do ensino superior a períodos de tempo e a salas [2]. É conhecida a lista de examesa marcar, divididos em grupos cuja calendarização deve ser espalhada o mais possível ao longo doperíodo de exames. Um exemplo típico de um destes grupos de calendarização são os exames dasunidades curriculares de um mesmo ano curricular, ou de uma especialização desse ciclo de estu-dos. São ainda conhecidas as inscrições dos estudantes nas várias unidades curriculares e, portanto,os exames que têm estudantes em comum, mesmo pertencendo a diferentes grupos de calendariza-ção.

Define-se T = {t1, .., tn } como o conjunto de períodos, que têm uma hora de início e duraçãoprecisas. Consideram-se 3 períodos diários, sem sobreposição.

Uma solução do problema consiste na atribuição de um período, t ∈ T , a cada exame, e ∈ E(E → T ). A restrição fundamental neste processo de atribuição é a não ocorrência de sobreposições,ou seja, para um par de exames que estão alocados no mesmo período ti ∈ T , ex e ey , recorrendoa uma função estudantes(ei ,e j ) que retorna o número de estudantes que estão inscritos a ambosos exames, deverá ocorrer estudantes(ex ,ey )= ;. O cumprimento desta restrição, permite obtervárias soluções admissíveis, que terão no entanto diferentes níveis de qualidade. A qualidade deuma solução pode ser avaliada usando diversos indicadores, como por exemplo: número de estu-dantes com exames em dias adjacentes ou o número de dias de intervalo mínimo para pares deexames consecutivos, entre outros.

Para tornar este problema adequado às necessidades do curso, foram definidas restrições fortese fracas. Exemplos de restrições fortes são que cada exame tem de ocorrer exatamente uma vez e quenum determinado período um estudante só possa ter no máximo um exame atribuído; exemplosde restrições fracas serão que o mesmo estudante não pode ter dois exames consecutivos com umintervalo inferior a n períodos ou exames com estudantes em comum não podem ser atribuídos aperíodos adjacentes.

IO 2017 75

ESTUDIO

Relevância do problemaEste problema tem sido abordado por vários investigadores [3], dada a sua dificuldade e relevânciaprática. Num caso particular de 80 exames a serem marcados ao longo de 10 dias, o número possívelde combinações atinge os 3080 ≈ 10118. A elaboração manual de um mapa de exames revela-seextremamente longa e com um resultado de fraca qualidade, surgindo, portanto, a necessidade deautomatizar este processo.

Metodologia utilizada na resoluçãoPara a resolução do problema vai ser usada a meta heurística BRKGA, biased random-key genetic al-gorithm, que é uma meta-heurística evolutiva baseada no algoritmo de chaves aleatórias, unbiased,de Bean, em 1994 [4]. Desenvolvido por Maurício Resende e J. Gonçalves [5], o BRKGA tem sido uti-lizado em problemas de otimização como o job-shop scheduling, multi-project scheduling, resourceconstrained project scheduling e outros, com excelentes resultados.

A codificação do cromossoma tem a seguinte estrutura: os genes correspondem aos examese o seu valor representa o período associado (um valor contínuo entre 0 e 1). É analisado cadacromossoma segundo as várias restrições que irão impingir penalizações afetando a função objetivo.A função objetivo é definida através de uma combinação linear das restrições fortes e fracas, em queas últimas apenas são contabilizadas se as primeiras forem cumpridas: f .o. = w1 · y( f or tes)+w2 ·δ ·z( f r acas), onde w1 e w2 são pesos e δ uma variável binária que toma o valor de 1 se as restriçõesfortes forem cumpridas e 0 se não.

ResultadosForam efetuadas análises, variando os vários parâmetros numa certa gama, num período de 3000gerações, os quais: número de populações, K, percentagem de cromossomas elite na população,pe ⊂ [0.1,0.3], percentagem de cromossomas mutantes na população, pm ⊂ [0.1,0.3], probabilidadede uma solução adquirir genes da solução elite, r hoe ⊂ [0.6,0.9], o intervalo de gerações em queas populações trocam entre si as melhores soluções (elite), X _I N T V L ⊂ [50,100], e esse mesmonúmero, X _NU MBER ⊂ [1,3].

Em todos os casos as restrições fortes foram cumpridas na sua totalidade devolvendo soluçõesadmissíveis com qualidade.

Referências[1] Nader Chmait and Khalil Challita. Using simulated annealing and ant-colony optimization

algorithms to solve the scheduling problem. Computer Science and Information Technology,1(3):208–224, 2013.

[2] Anthony Wren. Scheduling, timetabling and rostering - a special relationship? In Selected Papersfrom the First International Conference on Practice and Theory of Automated Timetabling, pages46–75, London, UK, UK, 1996. Springer-Verlag.

[3] R. Qu, E. K. Burke, B. McCollum, L. T. G. Merlot, and S. Y. Lee. A survey of search methodolo-gies and automated system development for examination timetabling. Journal of Scheduling,12(1):55–89, oct 2008.

[4] James C. Bean. Genetic algorithms and random keys for sequencing and optimization. ORSAJournal on Computing, 6(2):154–160, 1994.

76 IO 2017

ESTUDIO

[5] José Fernando Gonçalves and Mauricio G. C. Resende. Biased random-key genetic algorithmsfor combinatorial optimization. Journal of Heuristics, 17(5):487–525, 2011.

IO 2017 77

ESTUDIO

Optimal Control Strategies for an Advertisement ViralDiffusion

João N.C. Gonçalves1 Helena Sofia Rodrigues1 M. Teresa T. Monteiro1

1Algoritmi R&D Center, Department of Production and Systems, University of Minho, Braga,Portugal; Email: jncostagoncalves@gmail.com / tm@dps.uminho.pt

2School of Business Studies, Polytechnic Institute of Viana do Castelo, Valença, Portugal andCIDMA, Department of Mathematics, University of Aveiro, Aveiro, Portugal; Email:

sofiarodrigues@esce.ipvc.pt

The ProblemThe process of converting general advertisements into viral and attractive marketing messages re-quires major organizational efforts by the companies. In this context, traditional marketing strate-gies have struggled to meet demanding conditions of consumers [6]. Thus, recent research studies(see, e.g., [7]) have been trying to study marketing tactics to increase the levels of information diffu-sion within a target audience. One of these techniques is known as Viral Marketing (VM), that can beseen as an advertising strategy that benefits from the impact of network effects, mainly of word-of-mouth, to diffuse a marketing message into a large audience in a fast and economic way. However,some crucial questions remain:

1. Using VM strategies, how can the marketing message be optimally disseminated into the tar-get population?

2. When and what strategies should be applied in order to obtain the maximum levels of infor-mation spreading in an economically sustainable way?

Relevance of the ProblemSuccessful viral marketing campaigns depend on efective marketing strategies. At this point, shouldmarketers recruit susceptible population to spread a message or foster individuals who have alreadybeen in contact with it? By studying a real viral marketing campaign, optimal control strategies ap-plied in optimal time windows are compared in order to provide guidelines to maximize the spreadof information without losing money in future marketing advertisements.

Model ApproachConsidering that VM can be modeled by epidemiological models [8], our aim is to tackle the ques-tions stated before by analyzing a Susceptible-Infected-Recovered (SIR) epidemiological model withoptimal control theory, proposed in [9]. Using MATLAB software to solve the system of ordinary dif-ferential equations, numerical results were obtained using a forward-backward fourth-order Runge-Kutta’s scheme. Two optimal controls signals incorporated in the model, concerned the recruit-ment of targeted individuals to spread the message and the fostering of infected ones to continue tobroadly share it, are studied and compared under different investment cost scenarios. The ultimategoal is to meet a trade-off: maximize information spreading and minimize the costs associated to it.

ResultsResults suggest that the recruitment of susceptible population to act as spreaders has more impacton the cost functional than the strategy of fostering infected individuals. In general, optimal inter-vention strategies must be combined since the launching of the marketing campaign. It is shown

78 IO 2017

ESTUDIO

that the levels of information diffusion would have been higher if the company had implementedoptimal control theory. We also show that high investment costs in publicity strategies do not implyhigh overall levels of information diffusion. All in all, the benefits of using optimal control in theplanning and conception of future viral marketing campaigns are enhanced.

References[6] C. Pescher, P. Reichhart and M. Spann. Consumer Decision-making Processes in Mobile Viral

Marketing Campaigns. Journal of Interactive Marketing, 28(1):43–54, 2014.

[7] O. Koch and A. Benlian. Promotional Tactics for Online Viral Marketing Campaigns: HowScarcity and Personalization Affect Seed Stage Referrals. Journal of Interactive Marketing,32:37–52, 2015.

[8] J. Leskovec, L.A. Adamic and B.A. Huberman. The dynamics of viral marketing. ACM Trans. Web(TWEB), 1(1):1–39, 2007.

[9] K. Kandhway and J. Kuri. How to run a campaign: Optimal control of SIS and SIR informationepidemics. Applied Mathematics and Computation, 231:79–92, 2014.

IO 2017 79

ESTUDIO

Uniformização de consumos e centralização de compras demateriais clínicos em contexto hospitalar

Juliana Marques1 Nazaré Rego2

1Escola de Economia e Gestão, Universidade do Minho, Braga; Email:pg26812@alunos.uminho.pt

2Escola de Economia e Gestão, Universidade do Minho, Braga; Email: nazare@eeg.uminho.pt

Descrição do problemaDada a importância de reduzir custos nos sistemas logísticos e identificar possíveis fontes de melho-ria para fazer face ao agravado contexto económico, as organizações têm adotado estratégias de uni-formização dos consumos e de centralização das compras [10, 11]. O estudo desenvolvido analisouo impacto de uma situação potencial de uniformização e centralização de compras de um conjuntode hospitais (com gestão comum) nos principais custos de um sistema de gestão de inventários:custos de aquisição, custos de encomenda e custos de posse.

Relevância do problemaA centralização das compras, apoiada na partilha de informações através de tecnologias informaçãoe comunicação, permite obter um conjunto de sinergias, tais como economias de escala, de infor-mação e de processo [12]. As principais lacunas da literatura sobre este tema no contexto da saúderelacionam-se com a existência de poucos estudos empíricos sobre como podem ser obtidas melho-rias de eficiência através da centralização de compras e que incorporem processos de uniformizaçãoda informação sobre os artigos hospitalares.

Metodologia utilizada na resoluçãoA investigação realizada foi desenvolvida em contexto hospitalar - nas três unidades hospitalares daSanta Casa da Misericórdia do Porto (SCMP) e suportada pela metodologia de Estudo de Caso.

Procedeu-se à recolha e análise de dados quantitativos relativos às compras de dez meses de2015 das várias unidades hospitalares da SCMP para se analisar a estrutura de compras. Como acompra de artigos de Material de Consumo Clínico (MCC) representava cerca de 57% das comprasdas três unidades hospitalares, utilizaram-se 25 artigos de MCC consumidos pelas três unidadeshospitalares e 23 artigos de MCC de consumo regular e substituíveis (por exemplo, luvas do mesmotamanho apesar de poderem ter composição diferente), agrupados em 5 conjuntos de artigos semel-hantes, para quantificar o impacto nos custos de aquisição, encomenda e posse de inventáriosda uniformização dos consumos e da centralização das compras, tanto de artigos consumidospelas três unidades hospitalares, como de artigos semelhantes e substituíveis num só hospital. Asamostras recolhidas poderão ser consideradas ilustrativas de situações semelhantes noutros gruposde hospitais.

Após ter sido verificada a não normalidade da distribuições das procuras durante os prazos deentrega do fornecedor de cada um dos artigos analisados, aplicaram-se técnicas de simulação paragerar amostras plausíveis da procura conjunta diária dos vários artigos substituíveis, tendo em contaa procura diária (dados históricos) e as indicações de prazos de entrega do fornecedor obtidas dosserviços.

A análise do impacto da uniformização de consumos e centralização das compras de MCC nocusto de aquisição foi realizada por comparação entre o custo unitário a que o artigo foi adquirido

80 IO 2017

ESTUDIO

e seu o potencial custo unitário, caso o artigo tivesse sido comprado ao fornecedor ao qual, no con-junto das três unidades hospitalares, se comprou esse artigo (ou um que o pudesse substituir) aocusto unitário mais baixo.

Para analisar o impacto no inventário de segurança, utilizou-se o modelo estocástico de re-visão contínua e comparou-se os inventários de segurança necessários no caso de se satisfazer aprocura conjunta de artigos de consumo regular e substituíveis com os necessários considerando asprocuras individuais dos vários artigos.

O impacto da agregação das procuras dos artigos comuns às três unidades hospitalares nonúmero de encomendas foi determinado comparando o número de encomendas realizadas paraesses artigos pelas três unidades hospitalares (dados históricos) com o número de encomendasnecessárias, determinado a partir do modelo estocástico da revisão contínua, considerando a sim-ulação da procura conjunta diária desses artigos.

ResultadosA investigação realizada permitiu obter resultados de natureza quantitativa e qualitativa. Os re-sultados de natureza mais quantitativa mostram que a agregação das procuras de MCC nas trêsunidades hospitalares da Santa Casa da Misericórdia do Porto poderia originar uma poupança mé-dia de aproximadamente 11% ao nível dos custos de aquisição, de aproximadamente 30% em termosde inventário de segurança e uma redução de 55% no número de encomendas realizadas.

A investigação também permitiu concluir que a agregação das procuras dos vários hospitais e aexistência de uma gestão logística centralizada poderá originar poupanças de custos, economias deprocesso e de informação, um melhor aproveitamento dos recursos existentes com melhorias nosprocedimentos de trabalho e contribuir para a padronização da prática clínica.

Referências[10] U. Arnold. Purchasing consortia: Theoretical framework and empirical data. Revista de Econo-

mia e Direito, 1(2):5–26, 1996.

[11] K. Karjalainen. Estimating the cost effects of purchasing centralization–empirical evidencefrom framework agreements in the public sector. Journal of purchasing and supply manage-ment, 17(2):87–97, 2011.

[12] E. Tella and V.-M. Virolainen. Motives behind purchasing consortia. International Journal ofProduction Economics, 93:161–168, 2005.

IO 2017 81

Index

Plenary Speakers

Ana Paula Barbosa Póvoa . . . . . . . . . . . . . . . Page 15Ruth Kaufman FORS . . . . . . . . . . . . . . . . . . . . Page 16Ricardo Viana Vargas . . . . . . . . . . . . . . . . . . . .Page 16

Session Speakers

A

Aguiar, Ana Raquel. . . . . . . . . . . . .Wed.1.1, Page 19Almeida, João P. . . . . . . . . . . . . . . . . . . Fri.1.2, Page 57Alvelos, Filipe. . . . . . . . . . . . . . . . . . .Thu.1.4, Page 32Alvelos, Helena . . . . . . . . . . . . . . . . . Wed.1.4, Page 24Alves, Cláudio . . . . . . . . . . . . . . . . . . . . Fri.1.4, Page 60Alves, Maria João . . . . . . . . . . . . . . . Thu.1.1, Page 28Alves, Paula. . . . . . . . . . . . . . . . . . . . .Wed.1.3, Page 22Amaro, Ana Cristina . . . . . . . . . . . Wed.1.6, Page 26Aminbakhsh, Saman . . . . . . . . . . . Thu.1.3, Page 31Amorim, Pedro . . . . . . . . . . . . . . . . . Wed.1.6, Page 26Amorim Lopes, Mário . . . . . . . . . . . Fri.1.5, Page 61Andrade, Xavier . . . . . . . . . . . . . . . . Thu.3.4, Page 51Arantes, Amilcar. . . . . . . . . . . . . . . .Thu.2.3, Page 40Atashi Khoei, Arsham . . . . . . . . . . Thu.1.6, Page 35

B

Balsa, Carlos . . . . . . . . . . . . . . . . . . . . . Fri.1.3, Page 58Bandeira, Daniel . . . . . . . . . . . . . . . Wed.1.1, Page 19Barbosa Póvoa, Ana Paula . . . . . Wed.1.6, Page 26Basilio, Marcio . . . . . . . . . . . . . . . . . Thu.2.2, Page 38

C

Captivo, Maria Eugénia . . . . . . . . Thu.2.4, Page 42Cardoso, Domingos . . . . . . . . . . . . Thu.1.5, Page 34Cardoso-Grilo, Teresa . . . . . . . . . . Thu.2.4, Page 42Carravilla, Maria Antónia . . . . . . Thu.1.1, Page 28

Carvalho, Ana Sofia . . . . . . . . . . . . Thu.3.4, Page 51Carvalho, Margarida . . . . . . . . . . . Thu.1.5, Page 34Castilho Gomes, Marta . . . . . . . . . Thu.3.5, Page 53Castro, Pedro . . . . . . . . . . . . . . . . . . . . Fri.1.6, Page 63Chergui, Zhor . . . . . . . . . . . . . . . . . . . . Fri.1.2, Page 57Chibeles-Martins, Nelson . . . . . . . Fri.1.4, Page 60Cordeiro, Raphael . . . . . . . . . . . . . .Wed.1.1, Page 19Correia, José . . . . . . . . . . . . . . . . . . . . Thu.1.6, Page 35Correia Pinto, João. . . . . . . . . . . . .Wed.1.3, Page 22Costa, Ana Sara . . . . . . . . . . . . . . . . . Thu.1.2, Page 29Costa, Anabela. . . . . . . . . . . . . . . . . .Thu.3.5, Page 53Cruz-Gomes, Sofia . . . . . . . . . . . . . Thu.1.4, Page 32Cunha, Ricardo . . . . . . . . . . . . . . . . . . Fri.1.3, Page 58Cunha Ferreira, Diogo . . . . . . . . . Thu.1.4, Page 32Cunha Marques, Rui . . . . . . . . . . . Wed.1.2, Page 20Custodio, Ana Luisa . . . . . . . . . . . . Thu.2.5, Page 44

D

de Freitas, Jéssica. . . . . . . . . . . . . . .Thu.3.2, Page 49Dias, Luis . . . . . . . . . . . . . . . . . . . . . . . Thu.3.2, Page 49Dias, Luis C. . . . . . . . . . . . . . . . . . . . . Thu.1.2, Page 29Dias, Ruben . . . . . . . . . . . . . . . . . . . . . . Fri.1.5, Page 61

E

Esteves, C. Hiolanda . . . . . . . . . . . Wed.0.0, Page 19

F

Falcão e Cunha, Nuno . . . . . . . . . Wed.1.6, Page 26Faria, Joana . . . . . . . . . . . . . . . . . . . . . Thu.2.4, Page 42Fernandes, António . . . . . . . . . . . . Wed.1.2, Page 20Fernandes, Leão José . . . . . . . . . . . Thu.2.3, Page 40Fernandes, Susana . . . . . . . . . . . . . Thu.3.6, Page 54Figueira, Gonçalo . . . . . . . . . . . . . . Wed.1.4, Page 24Freitas, Liliana . . . . . . . . . . . . . . . . . . Thu.1.4, Page 32

IO 2017 83

INDEX

G

Gomes, Maria Isabel . . . . . . . . . . . . Thu.2.6, Page 46Gonçalves, João . . . . . . . . . . . . . . . . Wed.1.1, Page 19

I

Iglésias, Carlos . . . . . . . . . . . . . . . . . Thu.3.4, Page 51

K

Kostylenko, Olena . . . . . . . . . . . . . . . Fri.1.3, Page 58

L

Lima, Ricardo. . . . . . . . . . . . . . . . . . .Thu.3.3, Page 50Lopes, Isabel Cristina . . . . . . . . . . Wed.1.4, Page 24Lopes, Rui Borges. . . . . . . . . . . . . . .Thu.3.1, Page 47Lucas, Ricardo . . . . . . . . . . . . . . . . . . . Fri.1.5, Page 61

M

Machado Emiliano, William . . . Thu.3.3, Page 50Marques, Alexandra . . . . . . . . . . . . Thu.2.3, Page 40Marques, Catarina . . . . . . . . . . . . . . . Fri.1.6, Page 63Marques, Juliana . . . . . . . . . . . . . . . Wed.1.1, Page 19Marques, Rui . . . . . . . . . . . . . . . . . . . Wed.1.2, Page 20Martinho, Alberto . . . . . . . . . . . . . . Thu.3.2, Page 49Martins, Sara . . . . . . . . . . . . . . . . . . . Thu.3.1, Page 47Martins, Sara. . . . . . . . . . . . . . . . . . . . .Fri.1.2, Page 57Martins, Susana Rafaela . . . . . . . . Thu.2.4, Page 42Martins Fontes, Dalila B. M.. . . .Thu.1.2, Page 29Matos, Manuel. . . . . . . . . . . . . . . . . .Thu.1.2, Page 29Mehrsai, Afshin . . . . . . . . . . . . . . . . . Thu.3.1, Page 47Mendes, Jorge . . . . . . . . . . . . . . . . . . Wed.1.3, Page 22Miranda, João . . . . . . . . . . . . . . . . . . Thu.3.6, Page 54Monteiro, Marta . . . . . . . . . . . . . . . . Thu.2.4, Page 42Mota, Bruna . . . . . . . . . . . . . . . . . . . . Wed.1.6, Page 26Moura, Ana . . . . . . . . . . . . . . . . . . . . . Thu.1.6, Page 35

N

Neuenfeldt Júnior, Alvaro . . . . . Wed.1.4, Page 24Neves-Moreira, Fábio . . . . . . . . . . Thu.2.1, Page 37Nogueira, Mariana . . . . . . . . . . . . . . . Fri.1.1, Page 55

O

Oliveira, Beatriz Brito . . . . . . . . . . Thu.2.6, Page 46Oliveira, José . . . . . . . . . . . . . . . . . . . Wed.1.4, Page 24Oliveira, Larissa . . . . . . . . . . . . . . . . Thu.2.1, Page 37Oliveira, Max . . . . . . . . . . . . . . . . . . . Thu.2.2, Page 38

Oliveira, Monica. . . . . . . . . . . . . . . .Thu.2.2, Page 38Oliveira, Renata . . . . . . . . . . . . . . . . Wed.1.2, Page 20

P

P. Costa, M. Fernanda . . . . . . . . . . Thu.2.5, Page 44Pata, Arminda. . . . . . . . . . . . . . . . . . . .Fri.1.1, Page 55Paulo, Helena . . . . . . . . . . . . . . . . . . . . Fri.1.6, Page 63Pereira, Ana I. . . . . . . . . . . . . . . . . . . . . Fri.1.1, Page 55Pereira, Helga . . . . . . . . . . . . . . . . . . Thu.1.3, Page 31Pereira, Luís . . . . . . . . . . . . . . . . . . . . . Fri.1.5, Page 61Pereira, Sérgio . . . . . . . . . . . . . . . . . . Thu.2.5, Page 44Pinto, Telmo . . . . . . . . . . . . . . . . . . . . Thu.3.5, Page 53Pires, Maria . . . . . . . . . . . . . . . . . . . . Wed.1.3, Page 22

R

Ramos, António . . . . . . . . . . . . . . . . Thu.2.6, Page 46Rocha, Pedro . . . . . . . . . . . . . . . . . . . Thu.3.5, Page 53Rodrigues, Helena Sofia . . . . . . . . . Fri.1.3, Page 58

S

S. Pinto, Leonor . . . . . . . . . . . . . . . . Thu.2.6, Page 46S. Veloso, André . . . . . . . . . . . . . . . . Wed.1.2, Page 20Sadeghi, Parisa . . . . . . . . . . . . . . . . . . . Fri.1.4, Page 60Santos, Andreia . . . . . . . . . . . . . . . . . Thu.2.3, Page 40Santos, Nicolau . . . . . . . . . . . . . . . . . . Fri.1.1, Page 55Saylam, Serhat . . . . . . . . . . . . . . . . . Wed.1.3, Page 22Silva, Elsa . . . . . . . . . . . . . . . . . . . . . . . Thu.1.6, Page 35Silva, Sandra . . . . . . . . . . . . . . . . . . . . Thu.1.3, Page 31Silva Pinto, Francisco . . . . . . . . . . Thu.3.3, Page 50Soares, João. . . . . . . . . . . . . . . . . . . . .Thu.1.3, Page 31Soares, Ricardo . . . . . . . . . . . . . . . . . Thu.1.5, Page 34Soares de Morais, Carolina . . . . . Thu.2.6, Page 46Soeiro Ferreira, José. . . . . . . . . . . .Thu.3.6, Page 54

T

Tchemisova, Tatiana . . . . . . . . . . . Thu.2.5, Page 44Teixeira, Ana Paula . . . . . . . . . . . . . Thu.3.4, Page 51Teymourifar, Aydin . . . . . . . . . . . . . . Fri.1.2, Page 57Tsunoda Meira, William HitoshiThu.1.5, Page 34

V

Vaz, A.I.F. . . . . . . . . . . . . . . . . . . . . . . . .Thu.2.5, Page 44Vermuyten, Hendrik . . . . . . . . . . . Thu.3.1, Page 47Vieira, Ana . . . . . . . . . . . . . . . . . . . . . . Thu.2.2, Page 38Vieira, Manuel . . . . . . . . . . . . . . . . . . Thu.1.1, Page 28

84 IO 2017

INDEX

Vieira, Miguel . . . . . . . . . . . . . . . . . . . . Fri.1.6, Page 63

X

Xambre, Ana Raquel . . . . . . . . . . . .Thu.2.1, Page 37

Plenary Chairs

Ana Paula Barbosa Póvoa . . . . . . . A. Ismael F. VazRuth Kaufman FORS . . . . José Fernando OliveiraRicardo Viana Vargas . . . . . . . . . . . . . . . Lia Oliveira

Session Chairs

Wed.0.0, Page 19 . . . . . . . . . . . . . . . . . . . . Sandra SilvaWed.1.1, Page 19 . . . . . . . . . . . . . . . . . A. Ismael F. VazWed.1.2, Page 20 . . . . . . . . . . . . . . . . Clara Bento VazWed.1.3, Page 22 . . . . . . . . . . . . . . . . . . . . . Paula AlvesWed.1.4, Page 24 . . . . . . . . . . . . . . . . . Helena AlvelosWed.1.6, Page 26 . . . . . . Ana Paula Barbosa PóvoaThu.1.1, Page 28. . . . . . . . . . . . . . . .Maria João AlvesThu.1.2, Page 29 . . . . . . . . . . . . . . . . . . . . . Luis C. DiasThu.1.3, Page 31 . . . . . . . . . . . . . . . . . . . . Sandra SilvaThu.1.4, Page 32 . . . . . . . . . . . . . . . . . . . Filipe AlvelosThu.1.5, Page 34 . . . . . . . . . . . . . Domingos CardosoThu.1.6, Page 35. . . . . . . . . . . . . . . . . . . . . .Ana MouraThu.2.1, Page 37 . . . . . . . . . . . . Ana Raquel XambreThu.2.2, Page 38 . . . . . . . . . . . . . . . . Monica OliveiraThu.2.3, Page 40 . . . . . . . . . . . . . . . . Amilcar ArantesThu.2.4, Page 42 . . . . . . . . . Maria Eugénia CaptivoThu.2.5, Page 44. . . . . . . . . . . . .Ana Luisa CustodioThu.2.6, Page 46 . . . . . . . . . . . . Maria Isabel GomesThu.3.1, Page 47 . . . . . . . . . . . . . . . Rui Borges LopesThu.3.2, Page 49 . . . . . . . . . . . . . . . Alberto MartinhoThu.3.3, Page 50 . . . . . . . . . . . . . . . . . . . Ricardo LimaThu.3.4, Page 51 . . . . . . . . . . . . . . Ana Paula TeixeiraThu.3.5, Page 53. . . . . . . . . . . . . . . . . .Marta CastilhoThu.3.6, Page 54 . . . . . . . . . . . . . José Soeiro FerreiraFri.1.1, Page 55 . . . . . . . . . . . . . . . . . . . . Ana I. PereiraFri.1.2, Page 57 . . . . . . . . . . . . . . . . . . João P. AlmeidaFri.1.3, Page 58 . . . . . . . . . . Helena Sofia RodriguesFri.1.4, Page 60 . . . . . . . . . . . . . . . . . . . . Cláudio AlvesFri.1.5, Page 61 . . . . . . . . . . . . . . . . . . . . Ricardo LucasFri.1.6, Page 63 . . . . . . . . . . . . . . . Tânia Pinto-Varela

IO 2017 85

29

/jun

30

/jun

09:00Sh

uttle H

otel M

inh

o - H

otel Lara - ESC

E0

9:0

0Sh

uttle H

otel M

inh

o - H

otel Lara- ESC

E

10:50 - 1

1:20C

offee

Break

10:5

0 - 11

:20

Co

ffee B

reak

12

:00

12:2

0 - 13

:00

Aw

ard

s Cerem

on

y an

d C

losin

g Sessio

n

13

:00

Shu

ttle Ho

tel Min

ho

- Ho

tel Lara - ESCE

14

:00

16

:00

- 16

:20

16

:20

- 16

:50N

ote (1):

16:50 - 1

7:20C

offee

Break

18

:35

19:05Sh

uttle ESC

E - Ho

tel Lara - Ho

tel Min

ho

20:00Sh

uttle H

otel M

inh

o - H

otel Lara - "Q

uin

ta do

Am

orin

ho

"

22

:30

00:00Sh

uttle "Q

uin

ta do

Am

orin

ho

" - Ho

tel Lara - Ho

tel Min

ho

10

:45

(Arrival@

12

:00)

Shu

ttle Braga - V

alença (1

)

14

:30

(Arrival@

15

:45

)

17:20 - 1

9:00

20:30 - 0

0:00

14

:00

- 15

:00

15

:00

- 16

:00

16

:50

- 18

:30

19

:00

- 20

:00

20

:00

- 22

:30

Shu

ttle ESCE - M

uralh

as

Tou

r: Mu

ralhas d

e Valen

ça

Welco

me: "A

lvarinh

o d

e Ho

nra"

Shu

ttle Valen

ça - Ho

tel Min

ho

Plen

ary Session

III

Ricard

o V

argasP

arallel Session

s (five talks per sessio

n)

AP

DIO

Gen

eral A

ssemb

ly

Co

nferen

ce Din

ner

Lun

ch

28

/jun

Op

enin

g Sessio

n

Plen

ary Session

I

An

a Pau

la Pó

voa

Parallel Sessio

ns (five talks p

er session

)

Registratio

n

Shu

ttle Braga - V

alença

Parallel Sessio

ns (fo

ur talks p

er session

)P

arallel Session

s (fou

r talks per sessio

n)

Lun

ch

IO P

ro B

on

o: "R

ir é o M

elho

r Rem

édio

?"

Co

ffee B

reak

09:3

0 - 10

:50

09:30 - 1

0:50

15:30 - 1

6:50

A lu

ggage storage ro

om

will b

e available d

urin

g

the m

orn

ing o

f Jun

e 30.

11:2

0 - 12

:20

13:0

0 - 14

:30

11:20 - 1

3:00

13:00 - 1

4:30

14:30 - 1

5:30

Parallel Sessio

ns (fo

ur talks p

er session

)

Plen

ary Session

II

Ru

th K

aufm

an