Number 31 – 2019

ISSN Print 1414-8595 ISSN Online 2179-0655

NUMBER 31 – 2019

DIRECTORATE-GENERAL FOR NUCLEAR AND TECHNOLOGICAL DEVELOPMENT OF THE NAVY (DGDNTM)

NUMBER 31 – 2019 The mission of the Revista Pesquisa Naval is to provide a formal channel of communication and dissemination of national scientific and technical productions for the scientific community, by publishing original articles that are a result of scientific research and contribute to the advancement of knowledge in areas of interest for the Brazilian Navy. Articles published in the journal do not reflect the position or the doctrine of the Navy and are the sole responsibility of the authors.

SPONSORSHIP EDITORIAL BOARD Directorate-General for Nuclear and Technological Development of the CMG (T) Regina Boanerges Siqueira Navy – DGDNTM CF André Vinicius Faro Nunes CF Marco Aurélio de Castro Farias SPONSORSHIP SC Cristina Gonçalves Peroni Admiral Marcos Sampaio Olsen SC Janice Romaguera Trotte Duhá Director General for Nuclear and Technological Development of the Navy 2T (AA) Ana Cristina dos Santos Dutra EA Ernesto Carneiro Rodrigues ASSISTANT EDITORS Vice Admiral Noriaki Wada EDITION Director of the Navy Technological Center in São Paulo – CTMSP Directorate-General for Nuclear and Technological Development of Rear Admiral (EN) Luiz Carlos Delgado the Navy – DGDNTM Director of the Navy Technological Center in Rio de Janeiro – CTMRJ www.marinha.mil.br/dgdntm/revista Rear Admiral (RM1-EN) Humberto Moraes Ruivo Director of the Naval Agency for Nuclear Safety and Quality – AgNSNQ EDITORIAL PRODUCTION Rear Admiral (RM1) Paulo Cesar Demby Corrêa Zeppelini Publishers Chief Advisor of Institutional Relations & Social Communication – DGDNTM www.zeppelini.com.br

Revista Pesquisa Naval / Diretoria-Geral de Desenvolvimento Nuclear e Tecnológico da Marinha v. 1, n. 1, 1988 – Brasília – DF – Brasil – Marinha do Brasil

Anual Título abreviado: Pesq. Nav. ISSN Impresso 1414-8595 / ISSN Eletrônico 2179-0655

1. Marinha – Periódico – Pesquisa Cientiﬁca. Diretoria-Geral de Desenvolvimento Nuclear e Tecnológico da Marinha.

CDU 001.891.623/.9 CDD 623.807.2 CONTENTS | NUMBER 31 – 2019

FOREWORD

1 DIRECTOR-GENERAL OF NUCLEAR AND TECHNOLOGICAL DEVELOPMENT OF THE NAVY Fleet Admiral Marcos Sampaio Olsen COMPLEX SYSTEMS ANALYSIS

2 APPLICATION OF THE DISCRETE EVENT SIMULATION INTEGRATED WITH LEAN CONCEPTS FOR IMPLEMENTING IMPROVEMENTS IN A MILITARY FIELD HOSPITAL Aplicação da simulação de eventos discretos integrada com conceitos enxutos para a implementação de melhorias em um hospital militar de campo Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi MATERIALS DEVELOPMENT

14 CALIBRATION AND MODELING IN FINITE ELEMENTS OF PZT COMPACTION USING THE DRUCKER-PRAGER/CAP MODEL Calibração e modelagem em elementos ﬁnitos da compactação de PZT usando o modelo Drucker-Prager/cap Mateus M. Morais, Caiuã C. Melo, Rodrigo B. Canto, Carlos A. Fortulan

18 EFFECT OF ADDITIVES ON THE PREPARATION OF MEMBRANES FOR USE IN GAS-LIQUID CONTACTORS Efeito de aditivos no preparo de membranas visando o uso em contactores gás-líquido Gustavo Hidetaka Hashimoto, Cristina Cardoso Pereira, Felipe Brandão de Souza Mendes, Cristiano Piacsek Borges, Ana Paula Santiago De Falco

22 PRODUCTION OF GLYCEROL ETHERS ADDITIVES FOR IMPROVING COLD FLOW PROPERTIES Produção de aditivos à base de éteres do glicerol para melhoria de propriedades de ﬂuxo a frio Paula Araujo de Oliveira, Claudio José de Araujo Mota, Bianca Peres Pinto

26 SOL-GEL COATING CONTAINING ENCAPSULATED CORROSION INHIBITOR FOR PROTECTION OF AA2024 ALLOYS Revestimento sol-gel contendo inibidor de corrosão encapsulado para proteção de ligas AA2024 Lais de Campos Barbieri, Brenda Diane Pereira Martins, Renata Batista Rivero Garcia, Elizabete Yoshie Kawachi

31 BUILDING HIGH CAPACITY CYBER ARTIFACTS USING HIGH PRIVILEGED RESOURCES FROM MICROPROCESSORS Construção de artefatos cibernéticos de alta capacidade utilizando os recursos de alto privilégio de microprocessadores William Augusto Rodrigues de Souza OPERATIONAL RESEARCH

36 A PANORAMA OF THE USE OF OPERATIONAL RESEARCH IN THE SUBMARINE WAR ENVIRONMENT Um panorama do uso de pesquisa operacional no ambiente de guerra submarino Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense DECISION-MAKING PROCESS

43 A DECISION-MAKING SYSTEM FOR THE MARINE CORPS BASED ON MULTI-CRITERIA DECISION SUPPORT METHODS Uma sistemática de decisão para o corpo de fuzileiros navais baseada em métodos de apoio multicritério à decisão Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

54 USE OF THE SAPEVO-M METHOD WITH PARAMETERS OF THE SCOR 12.0 MODEL FOR RANKING OF SUPPLIERS IN A SUPPLY CHAIN OF HOSPITAL MATERIAL OF THE BRAZILIAN NAVY Utilização do método SAPEVO-M com parâmetros do modelo SCOR 12.0 para ranqueamento dos fornecedores em uma cadeia de suprimentos de material hospitalar da Marinha do Brasil Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos FOREWORD

Director-General of Nuclear and Technological Development of the Navy Fleet Admiral MARCOS SAMPAIO OLSEN

he perseverance and sustainability towards scienti!c, technological and innovation goals and achievements over the years have revealed signi!cant outputs on the frontiers of knowledge, most of them oriented to resolving challenges Tas imposed by societal needs. #e continuous search for optimization and overcoming of di$culties inherent to science requires coordinated e%orts that start with professional quali!cation and training, all the way down to scienti!c infrastructure and laboratory upgrad- ing, ultimately leading to disruptive technologies and complex systems under contemporary demands. In the !eld of National Defence, the Brazilian Navy, in tune with well recognized universities and industries in the country, and fully based on the principles of its own Strategy for Science, Technology and Innovation, has been gradually securing funds and essential investments for the promotion and the development of local and modern solutions targeted to dual applications under the banner of Naval Power to the needs and demands of local society. #e present edition of “Pesquisa Naval” provides scienti!c contributions in the !elds of operational research, complex system analyses, development of new materials and new decision-making processes. It is my duty to extend my congratulations to all authors presenting their !ndings and scienti!c achievements in this edition, in their quest for applied technology for the advancement of science within the Brazilian Navy. Enjoy your reading!

| 1 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 1 COMPLEX SYSTEMS ANALYSIS

APPLICATION OF THE DISCRETE EVENT SIMULATION INTEGRATED WITH LEAN CONCEPTS FOR IMPLEMENTING IMPROVEMENTS IN A MILITARY FIELD HOSPITAL Aplicação da simulação de eventos discretos integrada com conceitos enxutos para a implementação de melhorias em um hospital militar de campo

Carlos Henrique dos Santos 1, Gustavo Teodoro Gabriel 2, Afonso Teberga Campos 3, José Antonio de Queiroz 4, José Arnaldo Barra Montevechi 5

Abstract: Discrete Event Simulation (DES) and Lean Resumo: A Simulação de Eventos Discretos (SED) e o Lean apli- applied in health services help the quality and efficiency of cados em serviços de saúde auxiliam na qualidade e na e!ciência their services. The objective of this article is to use DES with dos serviços prestados. Diante disso, o objetivo do artigo é utilizar a the aid of Lean principles to analyze the structure of a mili- SED em conjunto com os princípios enxutos para analisar a estru- tary field hospital and to assess its ability to care for patients tura de um hospital militar de campo e avaliar a capacidade desse in case of environmental disasters. For this, the field hospi- hospital quanto ao atendimento de pacientes no caso de desastres tal was simulated using FlexSim Healthcare® software and, ambientais. Para isso, o hospital de campo foi simulado no software with the aid of Design of Experiments (DoE), the optimal FlexSim Healthcare ® e, juntamente com o planejamento de expe- number of locations and resources was defined. The main rimentos (DoE), foi de!nido o número ideal de locais e de recur- measure of output was the total number of patients com- sos. A principal medida de saída foi o número total de pacientes pletely treated. Results showed that, as currently configured, completamente tratados. Os resultados mostraram que, como está the field hospital is unable to cope with the demand, as only con!gurado atualmente, o hospital de campo não é capaz de lidar 20.6% of patients (588 out of 2,855) were completely trea- com a demanda, uma vez que apenas 20,6% dos pacientes (588 de ted. However, after changing the decision variables, the per- 2.855) foram completamente tratados. No entanto, após alterar as centage of fully treated patients increased to 93.6% (2673 variáveis de decisão, a porcentagem de pacientes completamente out of 2,855), which means an increase of 355%. To achieve tratados aumentou para 93,6% (2.673 de 2.855), signi!cando this result, the total number of locations increased by 162% aumento de 355%. Para alcançar tal resultado, o número total de (from 90 to 236), while the total number of human resources locais aumentou 162% (de 90 para 236), enquanto o número total decreased by 52% (from 63 to 30). de recursos humanos diminuiu 52% (de 63 para 30). Keywords: Discrete Event Simulation. Lean healthcare. Design of Palavras-chave: Simulação de Eventos Discretos. Lean Healthcare. Experiments. Field hospital. Planejamento de experimentos. Hospital de campo.

1. PhD student in Production Engineering, Universidade Federal de Itajubá – Itajubá, MG – Brazil. E-mail: [email protected] 2. PhD student in Production Engineering, Universidade Federal de Itajubá - Itajubá, MG – Brazil. E-mail: [email protected] 3. PhD student in Production Engineering, Universidade Federal de Itajubá - Itajubá, MG – Brazil. E-mail: [email protected] 4. Professor at Universidade Federal de Itajubá - Itajubá, MG – Brazil. E-mail: [email protected] 5. Professor at Universidade Federal de Itajubá - Itajubá, MG – Brazil. E-mail: [email protected]

| 2 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi

1. INTRODUCTION BARIL et al., 2017; WANG et al., 2015; HADDAD et al., 2016). For DES, integration allows focusing on Discrete Event Simulation (DES) is commonly used for efficiency gains by eliminating waste, using Lean con- studies and analysis of complex systems, such as manufac- cepts and techniques (RAGHAVAN et al., 2010; BHAT; turing and transportation systems, banks and even hospi- GIJO; JNANESH, 2014; YANG et al., 2015; HUANG ; tals (BANKS et al., 2010; SARGENT, 2013; LAW, 2015). KLASSEN; 2016). DES has played an important role since the 1960s when With regard to healthcare environments, there are dif- it comes to health care (BRAILSFORD et al., 2009) and, ferent types of hospitals, each designed to meet speci!c not long ago, hospitals and health authorities became more demands. Designed and managed by the United States aware of the potential of DES for improving the services Army, !eld hospitals (FH) are designed to serve people provided (CHENG; DRIVE; RABE, 2016). Examples of when natural disasters occur, as several hurricanes have hit studies that used DES in healthcare settings are found the North American continent in recent years. FH need to in the literature (HUSSEIN et al., 2017; BABASHOV provide immediate care to patients, and thus they need to et al., 2017; URIARTE et al., 2017). In addition, tools be well designed so that they can e$ciently cope with the common to manufacturing can also be applied to health- demand of patients. Hurricane-stricken areas often lose their care. Lean Healthcare can be implemented in the solu- sources of electricity and access to clean water, communication and in the optimization of processes in these areas tion, transportation, and distribution of food and medicine. (GRABAN, 2013). Many of these distribution channels remained inert and Lean applied in hospitals and health services can inoperable for several days after the disaster. For this reason, change very positively how these environments are every community must be prepared to provide immediate organized and managed. The Lean Healthcare meth- and lasting medical assistance to the populations a%ected odology provides hospitals with a shorter waiting time by these events. for patients, a reduction in errors made, and an increase In this context, the objectives of this study were to in the quality of services provided (GRABAN, 2013). analyze the structure of an FH and to evaluate its capac- Furthermore, the DES stands out as a promising tool ity and care of patients in the event of environmental in predicting behaviors of stochastic processes that disasters. In addition, the objective was also to evaluate have processing times. This method brings dynamism and define the ideal number of resources and equipment to the current state map and the map of the future state needed, as well as the most appropriate layout for efficient of the process, allowing for the inference on how resources service. Furthermore, the scope of the project aimed to are being used and where is there waste (DOGAN; enable the assistance of the largest possible number of UNUTULMAZ, 2014). patients without the resources being idle. For this, the Therefore, separately, DES and Lean have received scope of the study object was defined and the tools Lean increased attention in this area (ROBINSON et al., 2012). Healthcare, DES and Design of Experiments (DoE) Filser, Silva and Oliveira (2017) identified more than were applied together, to assess the current scenario and 300 studies on Lean in the health area, 99% of which possible improvements for future scenarios of the stud- were published from 2002. Brailsford et al. (2009) esti- ied hospital. mate that there are about 44,000 articles on simulation #e article is divided as follows: section 1 presents in health care, more than 80% published from the 1990s. the introduction and contextualization of the problem. However, when integrated, Lean and DES build a syner- Section 2 presents a literature review, de!ning DES and gistic relationship, leveraging its benefits. The possibility Lean Healthcare concepts and integration of these tools. of evaluating the results of the proposed improvements #en, section 3 presents the methodology used, de!ned in advance should be highlighted among the benefits of by Montevechi et al. (2007) and applied to the object of this integration for Lean, without the need to try them study, which is explained in section 4. Finally, in section in the real system (BHAT; GIJO; JNANESH, 2014; 5, the conclusions of the work are presented.

| 3 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi

2. BIBLIOGRAPHIC REVIEW #ese are the concepts of Lean Manufacturing, which when applied to healthcare systems are called Lean Healthcare. #e aforementioned wastes are adapted for hospital envi- 2.1. LEAN HEALTHCARE ronments (BERTRANI, 2012; GRABAN, 2013) and are According to Ohno (1997), the focus of Lean pro- explained according to Chart 1: duction is the elimination or minimization of waste, that is, eliminating any activity that does not add value 2.2. DISCRETE EVENT SIMULATION - DES to customers or users, but that adds cost to the product, #e simulation can be considered the imitation of a real whether in production or commercialization. Thus, activ- system, which is modeled on a computer for carrying out ities can be divided into three groups: activities that add later experiments to evaluate and improve the performance value, activities that do not add value, and activities that of this system (HARRELL et al., 2012). It can also be under- do not add value, but are necessary. Within the princi- stood as the creation and observation of a real or hypotheti- ples of Lean, Ohno (1997) classifies seven major wastes, cal system to generate inferences regarding it (BANKS et al., being: overproduction, stock, waiting, transportation, 2010). #is technique is extremely versatile and can be used handling, inefficient or unnecessary process, and defec- to investigate any type of stochastic system. Such versatil- tive processes. ity has made simulation the most used operational research

Chart 1. Comparison between Lean Manufacturing and Lean Healthcare. Waste Manufacturing Health services

Production above demand consumption Excessive monitoring in a patient who Overproduction capacity. It occurs by quantity does not demand such care. or by anticipation.

Excessive storage of goods. It harms customer service, causing waiting for Exam results awaiting processing, patients Storage information and products. May present waiting in line for care, expired materials transport and storage costs and and drugs that must be discarded. material damage.

Long periods of inactivity, such as waiting Waiting of the patient for release of Waiting for process, batch or operator, resulting in the bed, result of exams, treatments or long lead times. hospital discharge.

Unnecessary and/or excess transport of Excessive transportation of exams, Transportation goods and/or information, resulting in medications and patients that can be increased costs, time and enorts. caused by improper layout.

Excessive and/or unnecessary Excessive movement of doctors, movements of employees in the system. Movement nurses and assistants, due to poorly Include physical enorts during the organized jobs. manufacturing process.

Ineocient use of medication, surgical Ineocient or Procedures performed incorrectly; do not cart with missing item, ordering tests unnecessary process add value to the customer. that are not needed.

Generate products outside the customer’s Inadequate examinations, time/date data Defective processes speciﬁcation, either in product quality or posted on forms, but never used. delivery performance. Source: adapted from Bertrani (2012) and Graban (2013).

| 4 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi technique for studies dealing with random systems (HILLIER; to be carried out is the formulation of the problem. At this LIEBERMAN, 2001). For Law (2015) and Bloom!eld et al. stage, the process to be modeled is de!ned, so that the actions (2012), simulation is an alternative to direct experimentation and objectives can be speci!ed (BALCI, 2011). #e second in the real system, thus avoiding costs due to experimenta- stage is the construction and documentation of the conceptual tion and interrupting the ]ow of activities in the real system. model, followed by the validation of the model. Many tech- Among the advantages of the simulation, the following niques can be used for documentation, however, choosing a stand out: the decision rules and the possibility to explore technique focused on simulation is ideal (MONTEVECHI the information ]ows and the organizational procedures et al., 2007). #e last stage of the Conception is the modeling explored without interrupting the real system; to test phys- of the input data, which can be time, cost, percentages, capac- ical layouts and transportation systems without compro- ities, among others, varying according to the objective of each mising resources; to test the hypotheses as to the feasibil- study (BANKS et al., 2010; MONTEVECHI et al., 2007). ity of implementation; to expand or compress the time for #e second major phase, called Implementation, covers the investigation of phenomena; to study the importance the stages of construction, veri!cation, and validation of the of variables for system performance and answer “what-if ” computational model. #e construction of the computational questions (BANKS et al., 2010). #at is, in addition to promodel must be performed in software with which the modeler viding a complete view of the system, the simulation allows is familiar. #en, veri!cation is necessary, which will ensure decision makers to view the results even before their imple- that the programming of the computational model corre- mentation (LAW, 2015). sponds to the conceptual one (SARGENT, 2013). Finally, the validation of the computational model can be performed 2.3. LEAN HEALTHCARE AND through hypothesis tests, con!dence intervals, comparison DISCRETE EVENT SIMULATION charts, among others (SARGENT, 2013). #e integration of DES with Lean in health services assists #e last major phase, called Analysis, begins with the plan- in implementing Lean concepts on three major fronts: assess, ning, construction and analysis of the experiments. In this phase, facilitate and teach (ROBINSON et al., 2012). #us, the use possible scenarios are elaborated, in addition to using experi- of simulation and Lean Healthcare in hospital environments ment planning (DoE) and statistical tests (MONTGOMERY; can bring more safety, quality, and e$ciency for both patients RUNGER, 2018). After the experiments, the scenarios are and managers (GABA, 2004). In addition, this use can opti- analyzed, thus obtaining the conclusions and answers to the mize the ]ow of patients and serve as a motivational factor problem de!ned in the Conception stage. for employees (SALAM; KHAN, 2016). Swick et al. (2012) state that hospitals that integrate the two tools o%er an e$- cient method of strategic planning and provide employees 4. RESULTS AND DISCUSSION with a privileged view of how to reduce waste and add value. Still, it is possible to decrease the waiting time for patients, reducing the workload of employees and promoting the 4.1. CASE STUDY reallocation of resources (BHAT; GIJO; JNANESH, 2014; #e current FH consists of 12 tent units, which are cou- HADDAD et al., 2016). pled together, forming the structure of the hospital. It is estimated that the FH is capable of treating between 300 and 500 patients per day. In order to provide care, it is necessary 3. METHOD to pay attention to the di%erent types of patients, who can remain in the hospital for up to 48 hours. #us, it is assumed Modeling and Simulation, a technique used in the that patients receive !rst care at the FH facilities and, as soon study, can be divided into three major phases: Conception, as possible, are referred to the nearest health units. Implementation and Analysis (MONTEVECHI et al., 2007). Each patient who arrives at the FH goes straight into the In the !rst big phase, that is, the Conception, the !rst step Screening area. #ere, patients are assessed for their initial

| 5 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi condition and registered at the hospital. Patients are classi!ed health facility after 48 hours, with the help of an ambu- according to the severity of their injuries and then referred lance (15% of cases), or receives discharge and return to speci!c areas. If the patient’s severity is considered low, he home (85% of cases); is referred to the emergency room (ER) for super!cial care, • MA: the woman in labor is referred to an available bed, followed by discharged. If the patient’s severity is considered where she will undergo laboratory tests and the baby’s medium and there is a need for speci!c treatment, such as information will also be recorded. After the pre-delivery sutures, plaster, among others, they are sent to the ER, and time, the woman is taken to the delivery room and, after then taken to the appropriate sector. Finally, if the patient’s delivery, returns to the bed to recover, while the baby goes severity is high, they are directed to the speci!c care area, to the crib and waits for the mother’s recovery time to be which will depend on the type of care needed. taken to her. High severity patients can be referred to di%erent areas. • PS and PO: before surgery, the patient is submitted to If the patient is su%ering from a heart attack or other situation laboratory exams and X-rays, then waits until the time that requires immediate care and monitoring, the patient will of surgery in the PS area. #en, the patient goes into the be accompanied to the cardiac unit (CU) or to the extreme operating room (OR), where the procedure is performed. care ward (ECW). Women in labor are referred to the mater- At the end of the procedure, they are directed to the nity (MA) ward. Patients in need of surgical care are directed PO area to recover. After this period, the patient can be to the pre-surgical sector (PS) and after surgery, taken to the referred to the MCW (50.16% of the cases), the ICW postoperative sector (PO) for recovery. Patients who need (37.84% of the cases), and the ICU (12% of the cases); specialized doctors are referred to the minimal care (MCW) • ER: patients, upon arriving at the ER, wait in a wait- and intermediate care (ICW) wards. In these cases, patients ing chair until a medical assessment of their condition belonging to MCW are discharged after 24 hours, while those is carried out. After evaluation, they can be directed to belonging to ICW are discharged after 48 hours of observa- the suture area (about 10% of cases), where they are dis- tion. Finally, patients who are considered critical, but who do charged after the procedure. Patients can also be directed not !t the cases described, are referred to the ER. to the malaise (55% of cases) or to the infection treatment It is worth mentioning that all patients who are in the areas (20% of cases), where they are then directed to the speci!c care areas must be duly registered. If there are incon- MCW. Finally, patients can be referred to a psychological sistencies in the record, the admission and discharge team treatment area (3% of cases) or to a fracture area (12% of (AD) is responsible for regularizing such record in the bed cases), and subsequently taken to the ICW. itself; also, laboratory tests are performed. #e ]ow of each speci!c service area for cases of patients considered critical 4.2. CONCEPTION is described below: The study aimed to analyze the structure of a FH • CU and ECW: the patient, after arriving at the CU or the and to evaluate the capacity of that hospital regarding ECW, receives an initial assessment made by the group the care of patients in the case of environmental disas- of nurses, then undergoes regular assessment by nursing ters. Other issues to be resolved are: the ideal number of technicians until they are transferred to a health unit with employees and equipment, the most appropriate layout, and more resources. Such transfer is performed with air sup- the improvement of patient care according to the capac- port, depending on the risks to which the patient in this ity of human resources. #e construction and documen- condition is exposed; tation of the process that patients go through (described • ICW: the ]ow of the patient in this case is similar to the in section 4.1) was performed using IDEF-SIM (LEAL, ]ow of the CU and the ECW; however, the patient is 2008; MONTEVECHI et al., 2007). Figure 1 shows the referred to the nearest health unit after 24 hours, with conceptual model of the system. the help of an ambulance; The validation of the conceptual modeling was done • MCW: in this case, the ]ow is also similar to that of the through face-to-face validation, in which the specialists verify ICW; however, the patient can be referred to the nearest if the model in fact matches the real system. #e collection

| 6 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi and modeling of the data were made according to other nat- 4.4. ANALYSIS ural disasters that occurred on the continent. Historical data After the model was validated and simulated for a week, has undergone data processing, in which the best statistical it was possible to obtain the metrics for the !rst analyses. distributions have been linked. #erefore, it can be said that the hospital is unable to meet the expected demand. #us, the results obtained were: 4.3. IMPLEMENTATION • On average, 2,855 patients arrived at the model; #e computational model was built using the FlexSim • On average, 588 (20.6%) patients completed their treatments; Healthcare ® software. #is package was used due to its inter- • About 2,202 patients (77.01%) have not even been screened; face and ease of programming in relation to health services. • 65 patients (2.9%) started treatment, but did not !nish it. #e model was veri!ed by specialists. For validation, the model was simulated and validated using historical data. After a few executions, it was noted that human resources Figure 2 presents the screen of the current state of the com- were signi!cantly idle, indicating that they may not be a limit- putational model. ing feature. What can also be said for the layout characteristics

0.3% Recovery Baby Registration

Escorted to a Registration Group movement area Discharged MA & Laboratory (100%) & Childbirth & & Baby for mom Baby Crib Escorted Group Laboratory Group Escorted to a Obstetrician movement area Group Mov. Patient Time Before Giving Birth Escorted to a 37.84% MCW movement area Waiting for surgery 1 Group Mov. Patient and wheelchair OS Group 50.16% ICW Laboratory (100%) 2 3.8% OS/PO & & Surgery Postoperative & Group Mov. Patient Laboratory Group and wheelchair O 12.00% ECW Transported Group Mov. Transported to a Transported to a 3 Patient and Wheelchair X-Ray (20%) movement area Surgery movement area in a wheelchair Group in a wheelchair Group Mov. Patient Every 4 hours and wheelchair X-Ray group

Initial Evaluation Regular evaluation 2 Discharged accompanied Nursing Group Nursing Group 24 hours 2.5% ICW Patient Screening Registration x x & &

Intermediate Care Escorted Group Discharged Mov. Patient Screening Registration Every 1 hour Group Group Suture

Initial Evaluation Regular evaluation Discharged by Suture Group Nursing Group Nursing Group helicopter 0.6% CU 48 hours Illness treatment & & MCW x Cardiac Unit Nurses 1 Escorted Group Mov. Patient Infection treatment Escorted Group Mov. Patient Emergency Room

2.8% Nurses x Escorted to a Movement Area Psychological Medical evaluation x Escorted Group treatment Mov. Patient ICW 17.1% x Medical Group Nurses 2

Fractures Escorted Group Sent Discharged Mov. Patient 72.9% Emergency Room Group of Fractures

Screening Group

Every 4 hours Discharged Every 1 hour accompanied

Initial Evaluation Regular Evaluation Initial Evaluation Regular Evaluation 15% Discharged by helicopter Nursing group Nursing group Nursing group Nursing group 48 horas & x Discharged & & 1 3 Minimal care 85% Minimal care

Figure 1. Conceptual modeling of the simulated process.

| 7 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi and transport times. On the other hand, the MCW, OR, and Where: screening areas were occupied most of the time. In addition, Qi = Number of sites in area i; it was identi!ed that there is a long wait for assistance in the Pi = Mean daily number of patients arriving in area i; areas mentioned ti = Time (in hours) that the patient remains in area i. However, investigating the reasons why patients wait for care, a possible relationship between the areas was noted. Based on the results presented in Table 1, the variables #us, when fully occupied, the MCW blocks the predeces- OR, CU and beds for labor from the DoE were chosen to be sor areas, causing a sequential e%ect that returns to screening. excluded, since the ideal number needed for these locations is In fact, the number of MCW beds (19) is insu$cient in the similar to the current scenario. #us, the variables were rede!ned current state. On average, 67.3 patients require these beds per as shown in Table 2. #e central points were de!ned accord- day, and each patient remains there for approximately 48 hours. ing to the ideal number for each location (Table 1). A margin #us, about 135 beds are needed to satisfy the demand for this of 20% of the central points was used to de!ne “level -” and type of patient. However, this calculation considers only the mean “level +”. #e results were rounded up and down, respectively. estimate, which does not include the variability of the model #e minimum number of locations was established as one and the need for human resources in these experiments was elimi- 4.4.1. Analysis and experiments for locations nated, since they were idle in the !rst executions. DoE techniques were used to explore the variables A fractional factorial arrangement (fraction 1/8, resolu- related to the number of places (beds, tables, and chairs) in tion IV) was performed to screen the signi!cant variables, each area, since human resources and layout were not iden- resulting in 32 experiments. According to Figure 1, there ti!ed as the limiting characteristics. #e chosen variables are seven types of patients who follow seven di%erent ]ows. are shown in Table 1. #us, the number of patients seen by each ]ow was used #e numbers of required locations were determined in as a parameter for the DoE. After the analysis and based a deterministic manner in relation to the demand for each on the Pareto graphs generated for each DoE, variables B, area using Equation 1. #e results were rounded up to the F and G were eliminated, since they were not signi!cant next whole value and are also shown in Table 1. for the model. P * t Qi = i i (1) 24h Table 1. Variables related to the locations in each area. Variables Current Necessary

Beds (MCW) 19 135

Beds (ICW) 13 28

Beds (OR) 2 2

Beds (ECW) 4 2

Beds (CU) 3 2

Beds (PS) 8 3

Beds (ER) 5 2

Waiting room (ER) 8 16

Recovery beds (MA) 4 1

Beds for labor 1 1

Beds (Screening) 23 5

Total 90 197 Figure 2. Screen of the computational model: current state

| 8 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi

A response surface was de!ned using the centered face especially in the OR, PS, ER, and screening areas. #erefore, it experiment (CFC) arrangement, in which the numbers of was decided to increase the number of sites in these areas. locations have a minimum, maximum and a central point. In addition, other !nal adjustments were made, for which it Using this arrangement, it was possible to obtain 10 central was decided to change the variables referring to the beds of points and 10 axial ones, which resulted in another 20 exper- ICW, ECW, MA, and CU, according to Table 4. iments. It was decided to use this methodology to test the midpoints of each factor. Figure 3 shows one of the surfaces 4.4.2. Analysis and experiments for resources analyzed, which corresponds to decision variables D (PS beds) Based on the results obtained after the adjustments, human and E (ER beds) and the answer for the patient who needs resources were tested, as defined in the current scenario. intermediate care. In this case, it was observed that there is However, even after increasing the number of locations, employ- a curvature, indicating an optimum point. ees still found themselves stranded. #us, it was decided to For optimization, a multiobjective approach was adopted analyze the impact of these resources by reducing these groups by using the metamodels obtained in the DoE. #ese meta- through new simulation rounds. #e changes were made based models were combined using global criteria (RAO, 2009). on the resource occupation rate. One type of resource was ana- #e weights of each function were established using the lyzed at a time, starting with the most specialized. After a few canonical mixing polynomials method (TORRES et al., cycles, the changes were summarized and are shown in Table 5. 2016). #e optimal values of the variables and the results of the optimization for the one-week simulation are shown in Tables 3. #e values of variables B, F, and G were de!ned as Surface plot for moderate patients “level-”, since they were not considered signi!cant after the analysis of the DoE. After de!ning the ideal number of locations, the results 325 Hold Values obtained were: 300 A 135 • On average, 2,855 patients arrived at the model; 275 B 2 C 3 • On average, 2,674 (93.7%) patients completed their treatment; 250 • All patients (100.0%) underwent screening; 4 • 181 patients (6.3%) started treatment, but did not !nish it. 3 2 5 D E 1 6 Despite the improvements observed after the experiment, it is noted that patients still wait a long time for care, Figure 3. Response surface.

Table 2. Levels of design of experiment (DoE) variables for locations. Variables Description Level - Central point Level +

A Beds (MCW) 108 135 162

B Beds (ICW) 22 28 34

C Beds (ECW) 1 2 3

D Beds (PS) 2 3 4

E Beds (ER) 1 2 3

F Waiting chairs (ER) 12 16 20

G Recovery beds (MA) 1 2 3

H Beds (Screening) 4 5 6

| 9 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi

After de!ning the ideal number of resources and the mod- included. Table 6 presents the results of both the cur- el’s being simulated again for a week, it can be observed that: rent and the future scenario, illustrating that there was • On average, 2,855 patients arrived at the model; a reduction in the patient’s length of stay during the • On average, 2,673 (93.6%) patients completed their treatment; procedure (min). • All patients (100.0%) underwent screening; Figure 4 shows the layout for the future state of • 182 patients (6.3%) started treatment, but did not !nish it. the FH.

4.4.3. Conﬁrmation experiments Table 4. Number of locations in the current scenario To con!rm the proposed modi!cations, the model was and in the future. simulated 30 times over a week. According to the data observed Scenarios in Table 4, the total number of locations increased by 162.0%, Beds and chairs Current Future but, as shown in Table 5, the total number of human resources was reduced by 52.4%. Beds (Screening) 23 15 According to the simulation results, an increase in Recovery beds (MA) 4 2 waiting necessary in the ER (50.0%) and beds in the OR Waiting chairs (ER) 8 12 (50.0%), in the CU (66.7%), in the ICW (153.8%), and Beds (ER) 5 4 in MCW (700.0%). However, there was a need to reduce beds in screening, in MA, ER and PS by 34.8, 50, 20, Beds (PS) 8 5 and 37.5%, respectively. Regarding the ECW, there was Beds (OR) 2 3 no need for changes. As for human resources, changes Beds (ECW) 4 4 were not required for assistants, surgical technicians, and Beds (CU) 3 5 administrative technicians. For the other resources, there was a reduction of 20.0% for doctors, 42.3% for nurses, Beds (ICW) 13 33 91.6% for nursing technicians, 50.0% for X-ray techni- Beds (MCW) 19 152 cians, 66.7% for laboratory technicians, and 87.5% for Total 90 236 support technicians. For the final comparison between the current and the future state, in addition to the previously mentioned Table 5. Number of human resources in the current and future scenarios. metrics, the waiting time for each type of patient was Scenarios Human resources Current Future Table 3. Optimal values for the location variables. Doctors 5 4 Optimal Variables Description value Assistants 2 2

A Beds (MCW) 152 Nurses 26 15

B Beds (ICW) 22 Nursing technicians 12 1

C Beds (ECW) 3 Surgical technicians 2 2

D Beds (PS) 3 X-ray technicians 2 1

E Beds (ER) 2 Laboratory technicians 3 1

F Waiting chairs (ER) 12 Support technicians 8 1

G Recovery beds (MA) 1 Administrative technicians 3 3

H Beds (Screening) 5 Total 63 30

| 10 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi

5. CONCLUSION demand. Despite this, the proposed model provides adequate support for this need, since it has improved the service to #e present work aimed to use the integration of DES demand from 20.6 to 93.6%. All the questions proposed with lean concepts in a hospital environment. For this, the in the objective were answered and, using Lean principles, object of study chosen was a FH, which aimed to analyze there was a signi!cant improvement in the studied process. the structure and to evaluate the capacity of this hospital Regarding the limitations of this study, di$culties were regarding the care of patients in the case of environmen- found in the execution of the replicas due to the computa- tal disasters. tional e%ort required. For this reason, only one replica was #e Discrete Event Simulation methodology de!ned by performed in the current state model. In addition, the length Montevechi et al. (2007) was used in the study, which was of stay of patients was chosen as a measure of FH perfor- divided into three major phases: Conception, Implementation, mance, since it is the most critical metric from the patient’s and Analysis. In the Conception phase, the objectives were de!ned, namely: ideal number of employees and equipment; more appropriate layout and improvement of patient care according to the capacity of human resources. #e conceptual model was built using the IDEF-SIM modeling technique, and the input data was obtained using historical data from other situations in which a FH was needed. In the Implementation phase, the model was simulated in the FlexSim Healthcare ® software and also validated through historical data. For the Analysis phase, the DoE was used to verify the variables that were most in]uenced in the expanding locations. Experiments were used to determine the ideal number of human resources. #us, it can be said that the model in its current state does not have the capacity to meet the required Figure 4. Screen of the computational model: future state.

Table 6. Outputs for the current and future scenarios. Scenarios Outputs Current Future

Length of stay of pregnant women (MA) 0.2 13.1

Length of stay for surgical patients (PO) 5023.6 77

Length of stay for patients with minimal severity (MCW) 2073.4 22.9

Length of stay for cardiac patients (CU) 3140.5 12.5

Length of stay for patients with high severity (ECW) 5019.2 58.3

Length of stay for patients with intermediate severity (ICW) 5525.2 58.2

Length of stay of patients in the emergency room (ER) 2431.2 2.8

Total patients attended 588 2673

Total locations 90 236

Total human resources 63 30

| 11 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi point of view. However, having a high out]ow of patients 6. ACKNOWLEDGMENTS does not necessarily mean that they are being treated properly. #is is because the arrival of patients is irregular and The authors would like to thank the National Council the waiting time can be long. for Scientific and Technological Development (CNPq), Finally, for future work, it is suggested that di%erent the Coordination for the Improvement of Higher layouts are tested in order to assess the displacement of Education Personnel (CAPES), the Minas Gerais patients, human resources, and equipment. In addition, it State Research Support Foundation (FAPEMIG) is suggested to use the design of experiments to assess and and FlexSim for the support given to this and other optimize the amount of human resources and to assess researches. The authors also thank the SHS/FlexSim whether there is an in]uence on the interactions of these Model Building Competition for the second place resources with the di%erent changes to be made. acquired with the case presented.

REFERENCES

BABASHOV, V.; AIVAS, I.; BEGEN, M.A.; CAO, J.Q.; RODRIGUES, G.; of the academic literature on simulation and modelling in health D’SOUZA, D.; LOCK, M.; ZARIC, G. S. Reducing Patient Waiting Times care. Journal of Simulation , v. 3, n. 3, p. 130-140, 2009. https://doi. for Radiation Therapy and Improving the Treatment Planning Process: org/10.1057/jos.2009.10 a Discrete-event Simulation Model (Radiation Treatment Planning). CHENG, R.; DRIVE, E.; RABE, M. Simulation: the Past 10 Years and the Clinical Oncology, v. 29, n. 6, p. 385-391, 2017. https://doi.org/10.1016/j. Next 10 Years. In : WINTER SIMULATION CONFERENCE, 2016. Anais clon.2017.01.039 […]. Washington, D.C., 2016. BALCI, O. How to successfully conduct large-scale modeling and DOGAN, N.O.; UNUTULMAZ, O. Lean production in healthcare: a simulation projects. In : WINTER SIMULATION CONFERENCE, 2011. simulation-based value stream mapping in the physical therapy Anais […]. Phoenix, 2011. and rehabilitation department of a public hospital. Total Quality BANKS, J.; CARSON II, J.S.; NELSON, B.L.; NICOL, D.M. Discrete-event Management & Business Excellence , v. 27, n. 1-2, p. 64-80, 2014. Simulation. 5. ed. Nova Jersey: Pearson Prentice-Hall, 2010. 622 p. https://doi.org/10.1080/14783363.2014.945312

BARIL, C.; GASCON, V.; MILER, J.; BOUNHOL, C. The importance of FILSER, L.D.; SILVA, F.F.; OLIVEIRA, O.J. State of research and future considering resource’s tasks when modeling healthcare services with research tendencies in lean healthcare: a bibliometric analysis. discrete-event simulation: An approach using work sampling method. Scientometrics, v. 112, n. 2, p. 799-816, 2017. https://doi.org/10.1007/ Journal of Simulation , v. 11, n. 2, p. 103-114, 2017. https://doi.org/10.1057/ s11192-017-2409-8 jos.2016.6 GABA, D. The future vision of simulation in health care. Quality BERTRANI, T.M. Lean Healthcare: recommendations for and Safety in Health Care, v. 13, Supl. 1, p. 2-10, 2004. https://doi. implementations of the concepts of Lean Production in hospital org/10.1136/qhc.13.suppl_1.i2 environments . 166f. Tese (Doutorado) – Escola de Engenharia de São Carlos, Universidade de São Paulo, São Carlos, 2012. GRABAN, M. Hospitais Lean: melhorando a qualidade, a segurança dos pacientes e o envolvimento dos funcionários. 2. ed. Porto Alegre: BHAT, S.; GIJO, E.V.; JNANESH, N.A. Application of Lean Six Sigma Bookman, 2010. 312 p. methodology in the registration process of a hospital. International Journal of Productivity and Performance Management , v. 63, n. 5, p. HADDAD, M.G.; ZOUEIN, P.P.; SALEM, J.; OTAYEK, R. Case Study of 613-643, 2014. https://dx.doi.org/10.1108/IJPPM-11-2013-0191 Lean in Hospital Admissions to Inspire Culture Change. Engineering Management Journal, v. 28, n. 4, p. 209-223, 2016. https://doi.org/10.1 BLOOMFIELD, R.; MAZHARI, E.; HAWKINS, J.; SON, Y.J. Interoperability 080/10429247.2016.1234896 of manufacturing applications using the Core Manufacturing Simulation Data (CMSD) standard information model. Computer HARRELL, C.; GHOSH, B.K.; BOWDEN, R. Simulation using ProModel. & Industrial Engineering , v. 62, n. 4, p. 1065-1079, 2012. https://doi. 3. ed. Nova York: McGraw-Hill Education, 2012. 704 p. org/10.1016/j.cie.2011.12.034 HILLIER, F.S.; LIEBERMAN, G.J. Introduction to operations research . 7. BRAILSFORD, S.C.; HARPER, P.R.; PATEL, B.; PITT, M. An analysis ed. Nova York: McGraw-Hill, 2001. 976 p.

| 12 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 Carlos Henrique dos Santos, Gustavo Teodoro Gabriel, Afonso Teberga Campos, José Antonio de Queiroz, José Arnaldo Barra Montevechi

HUANG, Y.; KLASSEN, K.J. Using six sigma, lean, and simulation to Healthcare. European Journal of Operational Research , v. 219, n. 1, p. improve the phlebotomy process. Quality Management Journal, v. 23, 188-197, 2012. https://doi.org/10.1016/j.ejor.2011.12.029 n. 2, p. 6-21, 2016. https://doi.org/10.1080/10686967.2016.11918468 SALAM, M.A.; KHAN, S.A. Value creation through lean management: HUSSEIN, N.A.; ABDELMAGUID, T.F.; TAWFIK, B.S.; AHMED, N.G.S. a case study of healthcare service operations. International Journal Mitigating overcrowding in emergency departments using Six Sigma of Services and Operations Management , v. 25, n. 3, p. 275-293, 2016. and simulation: A case study in Egypt. Operations Research for Health https://doi.org/ 10.1504/IJSOM.2016.10000239

Care, v. 15, p. 1-12, 2017. https://doi.org/10.1016/j.orhc.2017.06.003 SARGENT, R.G. Veriﬁcation and validation of simulation models. LAW, A. Simulation modeling and analysis. 5. ed. Boston: McGraw-Hill Journal of Simulation , v. 7, n. 1, p. 12-24, 2013. https://doi.org/10.1057/ Education, 2015. 804 p. jos.2012.20

LEAL, F. Análise do efeito de falhas em processos de manufatura SWICK, M.; DOULAVERIS, P.; BAGNALL, T.; WOMACK, D. Application of Simulation Technology to Enhance the Role of the Professional através de projeto de experimentos simulados . 237f. Tese (Doutorado Nurse. The Journal of Nursing Administration , v. 42, n. 2, p. 95-102, em Engenharia de Produção) – Universidade Estadual Paulista “Júlio 2012. https://doi.org/10.1097/NNA.0b013e3182433706 de Mesquita Filho”, Guaratinguetá, 2008. TORRES, A.F.; GOMES, J.H.F.; CAMPOS, P.; PAIVA, A.P. Otimização MONTEVECHI, J.A.B.; PINHO, A.F.; LEAL, F.; MARINS, F.A.S. multiobjetivo com identiﬁcação de pesos ótimos: aplicação em Application of design of experiments on the simulation of a process um processo de torneamento de aço endurecido. In : SIMPÓSIO in an automotive industry. In : WINTER SIMULATION CONFERENCE, BRASILEIRO DE PESQUISA OPERACIONAL, 48., 2016. Anais [...]. 2007. Anais […], Washington, D.C., 2007. Vitória, 2016. MONTGOMERY, D.C.; RUNGER, G.C. Applied Statistics and Probability URIARTE, A.G.; ZÚÑIGA, E.R.; MORIS, M.U.; NG, A.H.C. How can for Engineers . 7. ed. Hoboken: Wiley, 2018. 976 p. decision makers be supported in the improvement of an emergency OHNO, T. O Sistema Toyota de Produção: Além da produção em department? A simulation, optimization and data mining approach. massa. Porto Alegre: Bookman, 1997. 150 p. Operations Research for Health Care , v. 15, p. 102-122, 2017. https://doi. org/10.1016/j.orhc.2017.10.003 RAGHAVAN, V.A.; VENKATADRI, V.; KESAVAKUMARAN, V.; WANG S.; KHASAWNEH, M.; SRIHARI, K. Reengineering the Cardiac WANG, T.; YANG, T.; YANG, C.; CHAN, F.T.S. Lean principles and Catheterization Lab processes: A lean approach. Journal of Healthcare simulation optimization for emergency department layout design. Industrial Management and Data Systems , v. 115, n. 4, p. 678-699, 2015. Engineering, v. 1, n. 1, p. 45-65, 2010. https://doi.org/10.1108/IMDS-10-2014-0296 RAO, S.S. Engineering Optimization: Theory and Practice. 4. ed. Nova YANG, T.; WANG, T.; LI, V.C.; SU, C.L. The Optimization of Total Jersey: John Wiley, 2009. 813 p. Laboratory Automation by Simulation of a Pull-Strategy. Journal ROBINSON, S.; RADNOR, Z.J.; BURGESS, N.; WORTHINGTON, C. of Medical Systems , v. 39, n. 1, p. 162, 2015. https://doi.org/10.1007/ SimLean: Utilising Simulation in the Implementation of Lean in s10916-014-0162-6

| 13 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 2-13 MATERIALS DEVELOPMENT

CALIBRATION AND MODELING IN FINITE ELEMENTS OF PZT COMPACTION USING THE DRUCKER-PRAGER/CAP MODEL Calibração e modelagem em elementos finitos da compactação de PZT usando o modelo Drucker-Prager/cap

Mateus M. Morais 1, Caiuã C. Melo 2, Rodrigo B. Canto 3, Carlos A. Fortulan 4

Abstract: #e design of molds for ceramic parts with complex Resumo: O projeto de moldes de peças cerâmicas com geometrias geometries presents challenges in predicting the dimensions and complexas apresenta desa!os quanto à previsibilidade da dimensão e density of the compacted part. Such di$culty can be overcome by da densidade da peça compactada. Tal di!culdade pode ser contor- using computer simulation with the !nite element method. #e nada pelo uso de simulação computacional por meio do método dos Drucker-Prager/cap constitutive model, used in simulations of elementos !nitos. O modelo constitutivo Drucker-Pra ger/ cap, usado powder materials, has a hardening curve that relates the hydros- em simulações de materiais em pó, utiliza curva de encruamento tatic pressure and the respective volumetric plastic strain. In order que relaciona a compressão hidrostática e a respectiva deformação to calibrate the hardening curve, disks of two batches of ceramic plástica volumétrica. Para calibrar essa curva, discos de dois lotes PZT powder were uniaxially pressed and then isostatically pres- de pós de cerâmica PZT foram pré-conformados uniaxialmente sed at seven pressure stages varying from 20 to 200 MPa. Model e, em seguida, foram prensados isostaticamente em sete patamares validation was performed by comparing !nite element simulation de pressão variando de 20 a 200 MPa. Foi realizada a validação do in ABAQUS software and experimental measurements, resulting modelo por meio de comparação entre simulação de elementos !ni- in maximum di%erences of 3.4% in dimension and 0.2% in den- tos no software ABAQUS e medidas experimentais, resultando em sity of compacted parts. Due to the great similarity of the results, diferenças máximas de 3,4% na dimensão e 0,2% na densidade das the calibrated model may be applied to more complex geometries. peças compactadas. Em função da boa aproximação dos resultados, Keywords: Finite elements. Lead zirconate titanate. Powder o modelo calibrado poderá ser aplicado em geometrias complexas. pressing. Palavras-chave: Elementos !nitos. Titanato zirconato de chumbo. Prensagem de pó.

1. Master in Mechanical Engineering from São Carlos School of Engineering, Universidade de São Paulo. PhD student in Mechanical Engineering at São Carlos School of Engineering, Universidade de São Paulo – São Carlos, SP – Brazil. E-mail: [email protected] 2. Master in Materials Engineering from Universidade Federal de São Carlos. PhD student in Materials Engineering at Universidade Federal de São Carlos – São Carlos, SP – Brazil. E-mail: [email protected] 3. PhD in Mechanical Engineering from São Carlos School of Engineering, Universidade de São Paulo and École Normale Superiéure de Paris. Adjunct Professor in the Materials Engineering Department at Universidade Federal de São Carlos – São Carlos, SP – Brazil. E-mail: [email protected] 4. PhD in Mechanical Engineering from the São Carlos School of Engineering, Universidade de São Paulo. Associate Professor in the Department of Mechanical Engineering, São Carlos School of Engineering, Universidade de São Paulo – São Carlos, SP – Brazil. E-mail: [email protected]

| 14 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 14-17 Mateus M. Morais, Caiuã C. Melo, Rodrigo B. Canto, Carlos A. Fortulan

1. INTRODUCTION 2. MATERIALS AND METHODS

Designing molds for pressing ceramic parts with com- For the compaction tests, two batches of PZT (from plex geometries is not simple, especially in applications that manufacturer Sparkler Piezoceramic Powder, grade: SP-4, require precise geometry and homogeneous density, such as without binder) were prepared by adding 0.8% by weight hydrophones and other types of sensors and piezoelectric of polyvinyl butyral (PVB) as a binder. Both batches actuators. Trial and error mold design can be very costly and (named A and B) were ground for 24 hours in a ball ine$cient. In this context, numerical simulation using the mill. Batch B was subsequently ground for 96 hours in !nite element method (FEM) is a useful tool, as it allows a a vibrating mill. Batch A’s weight is 50% composed of better prediction of the part’s dimensions after pressing, as particles smaller than 3.1 µm, whereas for batch B, this well as analysis of the density distribution in the part, show- value is 0.85 µm. ing critical regions that are more susceptible to cracks or to #e isostatic pressing test was used to calibrate the hard- causing deformations during sintering, which makes it pos- ening of the constitutive model for PZT material. Each batch sible to obtain a higher quality part with fewer iterations of powder was uniaxially preformed in a 41 mm diameter (MORAIS et al., 2016; CUNNINGHAM; LAMARCHE; mold with axial tension of 10 MPa and then was isostatically ZAVALIANGOS, 2017). pressed at the following pressures: 20, 40, 60, 80, 120, 160, To perform the simulation, it is necessary to identify a and 200 MPa. After each pressing stage, the dimensions of constitutive model for each material involved in the press- the tablets were measured, and the density was calculated ing and calibrate the parameters of these models exper- using the initial mass of each tablet. #e relationship between imentally (MORAIS et al., 2016). For the simulation of pressure and macroscopic volumetric plastic deformation was ceramic powders, such as PZT, the Drucker-Prager/cap adjusted by an exponential !t. #e other parameters necessary model is suitable, as it is used in compaction simulations for the simulation were adopted based on the literature on for granular materials, such as soils, pharmaceutical tab- alumina (MELO et al., 2018). Despite having been obtained let powders and metallic powders (HAN et al., 2008; from another ceramic material, these data have less in]uence DASSAULT SYSTÈMES, 2014; ZHOU, 2017). #ree than the hardening curve on the result of the simulation of surfaces de!ne the model: the Drucker-Prager surface, the isostatic pressing. which de!nes the ]ow according to the shear, the cap sur- For validation, a simulation of the compaction test of face that models an inelastic hardening mechanism, repre- batches A and B was performed in the standard solver of the senting the plastic nature of the compaction, and a transi- ABAQUS software. Due to the axial symmetry of the problem, tion surface that smoothly joins the previous two segments to facilitate numerical implementation. #e equations that de!ne these surfaces are functions of the von Mises stress Transition surface q q and the pressure p. (DASSAULT SYSTÈMES, 2014). Cap In Figure 1, the Drucker-Prager/cap model is illustrated Drucker-Prager surface surface in the diagram q vs. p. In addition to the constants that determine each of the three surfaces, it is necessary to obtain experimentally the hardening E curve of the cap surface that relates the stress of the hydro- Mises stress, Von static compression of the ]ow and the respective volumetric d p p plastic deformation (HAN et al., 2008; MELO et al., 2018). a b #is study aimed to describe the experimental calibration Hydrostatic pressure, p of the hardening curve of the Drucker-Prager/cap model of Source: adapted from Morais et al. (2016) and Dassault two batches of PZT and to validate this model through the Systèmes (2014). !nite element simulation in the ABAQUS software. Figure 1. Drucker-Prager/cap model.

| 15 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 14-17 Mateus M. Morais, Caiuã C. Melo, Rodrigo B. Canto, Carlos A. Fortulan the modeling is represented in two dimensions through the 3. RESULTS AND CONCLUSION revolution surface of the disk’s cross-section. #e mesh consisted of !rst-order continuous four-node axisymmetric ele- Figure 3 shows the density pro!le of the axisymmetric sec- ments with reduced integration (CAX4R). #e mold walls tion of the batch B disk obtained in the simulation. Table 1 shows and piston surfaces were modeled as rigid surfaces, with a the comparison between the dimensions and densities mea- contact friction coe$cient of 0.3. #e simulation steps are sured in the compaction test and the values obtained by the outlined in Figure 2. simulation. #e granulometry of the powders in]uenced the

1- Initial stage: 2- Uniaxial 3- Removal 4- Removal 5- Removal of 6- Removal 7- Isostatic 8- Total application of the pressing: of uniaxial of the upper the mold wall. of the lower pressing: depressurization. boundary removal of the force. punch. punch. applying conditions (axis of y-displacement 200 MPa symmetry, ﬁxed restriction of the pressure to lower punch, upper punch and the surface upper punch, and application of a of the piece. wall) and contact normal force of (powder with 19.8kN to the mold). upper punch. Upper punch 19.8 kN 200 Mpa Powder Mold wall Lower Symmetry axis Symmetry punch

Figure 2. Finite element simulation steps.

Density (g/cm 3) 4.136 3.984 3.832 3.680 3.528 3.376 3.224 Uniaxial 3.072 2.920 2.768 2.616 2.464 2.312

Density (g/cm 3) 4.936 4.928 4.920 4.912 4.904 4.896 4.887 4.879 4.871 4.863 Isostatic 200 MPa Isostatic 4.885 4.847 4.838

Figure 3. Density profiles of the axisymmetric cross-section of the batch B disk simulated by finite elements. Results of the uniaxial and isostatic pressing steps. Axial symmetry axis in red.

| 16 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 14-17 Mateus M. Morais, Caiuã C. Melo, Rodrigo B. Canto, Carlos A. Fortulan

Table 1. Comparison between computer simulation data and experimental tests in batches A and B, after the final isostatic pressing (200 MPa). Batch A Batch B Computer Difference Computer Difference Experimental Experimental simulation (%) simulation (%) Heights (mm) 6.3 6.4 1.6 5.8 5.6 -3.4 Diameter (mm) 38.4 38.5 0.3 36.6 37.3 -0.8 Density (g/cm 3) 5.20 5.19 -0.2 4.91 4.92 0.2

compaction of the pieces, which is evidenced by the di%erent This study was financed in part by the Coordenação densities between the batches. #is calibration methodology de Aperfeiçoamento de Pessoal de Nível Superior – Brasil showed an excellent approximation between the experimen- (CAPES) – Finance Code 001, and by the grants 2018/07928-2 tal and numerical results in this simple geometry. Due to this and 2018/02801-4, São Paulo Research Foundation (FAPESP), favorable result, the calibrated model can be used preliminarily and grant 131685/2018-0, National Council for Technological in future works to simulate more complex geometries, such and Scienti!c Development (CNPq). #e opinions, hypotheses, as hydrophones or other types of PZT sensors and actuators. conclusions, or recommendations expressed in this material #e results can be improved through other mechanical tests are the responsibility of the authors and do not necessarily to calibrate other parameters of the constitutive model. re]ect the views of these agencies.

REFERENCES

CUNNINGHAM, J.; LAMARCHE, K.; ZAVALIANGOS, A. Modeling MELO, C. C.; MORAES, A.L.I.; ROCCO, F.O.; MONTILHA, F.S.; CANTO, of powder compaction with the drucker-prager cap model. In : R.B. A Validation Procedure for Numerical Models of Ceramic Powder PANDEY, P.; BHARADWAJ, R. (orgs.). Predictive Modeling of Pressing. Journal of the European Ceramic Society , v. 38, n. 8, Pharmaceutical Unit Operations . Cambridge Woodhead Publishing, p. 2928-2936, 2018. https://doi.org/10.1016/j.jeurceramsoc.2018.01.009 2017. p. 205-227. MORAIS, M.M.; FORTULAN, C.A.; PURQUERIO, B.M.; PEDROSO, M.P.G. DASSAULT SYSTÈMES. Abaqus 6.14 Online Documentation - Abaqus Modelagem da densificação de esferas cerâmicas verdes prensadas Analysis User’s Guide . Dassault Systèmes, 2014. isostaticamente. In : CONGRESSO BRASILEIRO DE CERÂMICA, 60., 2016, Águas de Lindoia. Anais [...]. Águas de Lindoia, 2016. HAN, L.H.; ELLIOTT, J.A.; BENTHAM, A.C.; MILLS, A.; AMIDON, G.E.; HANCOCK, B.C. A Modified Drucker-Prager Cap Model for Die ZHOU, M.; HUANG, S.; HU, J.; LEI, Y.; XIAO, Y.; LI, B.; YAN, S.; ZOU, F. A Density- Compaction Simulation of Pharmaceutical Powders. International Dependent Modified Drucker-Prager Cap Model for Die Compaction Journal of Solids and Structures , v. 45, n. 10, p. 3088-3106, 2008. of Ag57.6-Cu22.4-Sn10-In10 Mixed Metal Powders. Powder Technology , https://doi.org/10.1016/j.ijsolstr.2008.01.024 v. 305, p. 183-196, 2017. https://doi.org/10.1016/j.powtec.2016.09.061

| 17 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 14-17 MATERIALS DEVELOPMENT

EFFECT OF ADDITIVES ON THE PREPARATION OF MEMBRANES FOR USE IN GAS-LIQUID CONTACTORS Efeito de aditivos no preparo de membranas visando o uso em contactores gás-líquido

Gustavo Hidetaka Hashimoto 1, Cristina Cardoso Pereira 2, Felipe Brandão de Souza Mendes 3, Cristiano Piacsek Borges 4, Ana Paula Santiago De Falco 5

Abstract: Gas liquid membrane contactors have been applied as Resumo: Membranas como contactores gás-líquido têm sido uti- an alternative to traditional gas separation processes due to the lizadas como alternativa aos processos convencionais de separação large interfacial area, avoiding direct contact between phases, de gases, pois promovem elevada área de contato, assim como evi- which leads to operational problems such as foaming. #is work tam o contato direto entre as fases e consequentes problemas ope- aims to evaluate the e%ect of additives in the formation of poly- racionais, como a formação de espumas. Este trabalho visa avaliar vinylidene ]uoride membranes (PVDF). Adipic acid (AD), activa- o efeito de aditivos no preparo de membranas de poli(]uoreto de ted carbon (AC) and water were used as additives in di%erent con- vinilideno) (PVDF). Foram utilizados como aditivos ácido adípico centrations. #e results of precipitation velocity indicated that the (AD), carvão ativado (CA) e água, em diferentes concentrações. higher AD concentration has probably provided viscosity increase, Os resultados de velocidade de precipitação indicam que a maior which inhibits non-solvent in]ow to polymer solution, leading to concentração de AD provavelmente promoveu aumento na viscosi- precipitation delay. Membrane morphologies, gas permeation and dade da solução, di!cultando a entrada de não-solvente na solução contact angle results have pointed out that the membranes present polimérica, levando à precipitação mais lenta da solução polimé- adequate permeability and hydrophobicity for use in membrane rica. As morfologias das membranas obtidas e os resultados de per- contactors processes. meação de gás e ângulo de contato também mostram que as mem- Keywords: Membranes contactors. Gas-liquid separation. branas apresentam valores de permeabilidade e hidrofobicidade polyvinylidene ]uoride membranes hollow !bers. adequados para uso nos processos de contactores com membranas. Palavras-chave: Contactores com membranas. Separação gás- líquido. Fibras ocas de poli(]uoreto de vinilideno).

1. Undergraduate Student in Chemical Engineering by the School of Chemistry at Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 2. D.Sc. in Chemical Engineering by Universidade Federal do Rio de Janeiro. Researcher at the Laboratory of Membrane Processes at Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 3. Master in Nanotechnology Engineering by Universidade Federal do Rio de Janeiro . In charge of the Chemical Processes Division of the Navy Research Institute – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 4. D.Sc. in Chemical Engineering by Universidade Federal do Rio de Janeiro. Associate Professor at Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 5. D.Sc. in Science and Technology of Polymers by the Institute of Macromolecules Professor Eloisa Mano at Universidade Federal do Rio de Janeiro. In charge of the Materials Technology Group at the Navy Research Institute – Rio de Janeiro, RJ – Brazil. E-mail: [email protected]

| 18 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 18-21 Gustavo Hidetaka Hashimoto, Cristina Cardoso Pereira, Felipe Brandão de Souza Mendes, Cristiano Piacsek Borges, Ana Paula Santiago De Falco

1. INTRODUCTION Polymer solutions with di%erent additives were used in the preparation of ]at membranes to better elucidate the phase

#e removal of carbon dioxide (CO 2) from gas streams inversion process. Subsequently, one of the conditions was is attractive because, in addition to being one of the gases selected to prepare the membrane in the form of a hollow responsible for the greenhouse e%ect, its presence can cause !ber. Light transmittance tests were performed to verify the corrosion of the ducts that transport it and reduce the cal- precipitation speed of the polymer solutions. #e solutions ori!c value of the natural gas. Conventional processes used were spread on glass plates and then immersed in the non-sol- to remove CO 2 include absorption with amines and cryo- vent bath. #en, data on decreasing light transmittance over genics. Although these processes are widely used, they have time were collected using a recorder. #e membranes obtained a number of disadvantages, such as high cost and complex were evaluated using scanning electron microscopy (SEM) operation. #e separation processes with membranes (PSM) (FEI Company), contact angle measurements (Dataphysics), have been receiving great prominence, as they bring advan- and gas permeability. tages that allow them to compete with traditional processes (ZHANG, WANG, 2014). In the case of the contactor process, the membranes do 3. RESULTS not act as selective barriers as occurs in traditional PSM. #e membranes are placed between two phases, e.g. gaseous Figures 1A and 1B show the results of the light transmit- and liquid, which promotes a wide contact area for mass tance tests, done to investigate the e%ect of the variation of the transfer, without dispersion between the phases. #e use of additive concentration and temperature on the precipitation membranes makes the interfacial area known and constant, speed of the polymer solution, respectively. Figure 2 shows making it easier to determine its performance when com- the e%ect of the concentration of water and activated car- pared to other processes such as the bubble column. bon in the solution. Polyvinylidene ]uoride (PVDF) is a polymer of interest Comparing the polymer solutions containing 5% wt in the preparation of membranes as gas-liquid contactors, of AD, it is observed that the increase in temperature due to its hydrophobicity and chemical resistance. In this increased the time for the beginning of precipitation, indi- work, the in]uence of additives on the formation of PVDF cating that such increase made the polymer solution more membranes is investigated for the preparation of hollow stable. #e preparation of polymer solutions with a higher !bers and for the use of these membranes in gas-liquid concentration of AD was only possible with increasing contactors, aiming at the removal of CO 2 from gas streams temperature. #is result indicates that the increase in AD (KANG, CAO, 2014). concentration leads to instability of the polymer solution and consequently to phase separation. #is result is con- !rmed by observing that the precipitation rate of the 7% 2. EXPERIMENTAL METHODOLOGY wt solution is faster than the precipitation of the 5% wt AD solution. In the case of the solution containing 10% PVDF membranes (Kynar 740, Arkema) were obtained wt AD, a longer time to start the precipitation of the poly- by the process of phase inversion by immersion – precipi- mer solution is observed, which is probably due to the tation, using solutions of PVDF, 17% wt, and N-methyl-2- increase in the viscosity of the polymer solution, making pyrrolidone (NMP) as solvent. Adipic acid (AD), activated it di$cult for the non-solvent to enter the bath (water) carbon (AC) and distilled water were used as additives in the in the polymer solution. polymer solutions. NMP and AC were purchased from Isofar. The solutions containing 5% AD with 1% water, and AD was purchased from Riedel-de-Haen. AD concentrations 5% AD with 1% activated carbon showed faster precipita- varied from 5 to 10 % wt. AC was added to 1% of the poly- tion speeds between the investigated membranes, depend- mer mass in the polymer solution, while the concentration ing on the approach of the solutions to the phase sepa- of distilled water varied from 0.5 to 1.0 % wt. ration region. The photomicrographs shown in Figure 3

| 19 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 18-21 Gustavo Hidetaka Hashimoto, Cristina Cardoso Pereira, Felipe Brandão de Souza Mendes, Cristiano Piacsek Borges, Ana Paula Santiago De Falco show that the obtained morphologies are in accordance 5. ACKNOWLEDGMENTS with the precipitation speed results, indicating that the faster precipitation at the interface of the polymer solu- #e authors would like to thank the National Council for tion with the precipitation bath provides resistance to Scienti!c and Technological Development for the !nancial mass transfer, which leads to an increase of macropores support and scholarship of Gustavo Hashimoto. in the sublayer. This effect is also observed through the cross section of the hollow fiber, in which immersion in the water bath promotes the increase of macropores from 100 outside to inside the fiber. Figure 4 shows that the solu- 80 tions containing AD at room temperature and addition 60 of water and AC provided an increase in the speed of 40 precipitation sufficient to generate a sublayer with reg- 20 ular macropores. Transmittance (%) Transmittance #e nitrogen permeation results showed an average value 0 0 100 200 300 400 500 600 of 1.5 × 105 GPU and a contact angle of 78º. Time (s)

5% AD (60°C) 4. CONCLUSIONS 5% AD + 1% water (60°C) 5% AD + 1% activated carbon (60°C)

#e e%ect of the additives was evident in the precipitation PVDF: Polyvinylidene ﬂuoride; AD: Adipic acid; speed tests of the polymer solutions. #e obtained morpholo- NMP: N-Methyl-2-pyrrolidone. gies and the results of gas permeation and contact angle also Figure 2. Light transmittance tests. (A) Effect of water and show that the membranes have adequate permeability and activated carbon concentration in the polymer solution of hydrophobicity values for use in contactors. PVDF/AD/NMP. PVDF = 17% wt and AD = 5% wt.

A B 100 100 80 80 60 60 40 40 20 20

Transmittance (%) Transmittance 0 (%) Transmittance 0 0 100 200 300 400 500 600 0 100 200 300 400 500 600 Time (s) Time (s)

5% AD (60°C) 5% AD (23°C) 7% AD (60°C) 5% AD (60°C) 10% AD (60°C)

PVDF: Polyvinylidene ﬂuoride; AD: Adipic acid; NMP: N-Methyl-2-pyrrolidone. Figure 1. Light transmittance tests. (A) Effect of the additive concentration in the polymer solution of PVDF/AD/ NMP. PVDF = 17% wt. Temperature of the polymer solution = 60°C. (B) Effect of the temperature of the polymer solution PVDF/AD/NMP = 17/5/78% wt.

| 20 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 18-21 Gustavo Hidetaka Hashimoto, Cristina Cardoso Pereira, Felipe Brandão de Souza Mendes, Cristiano Piacsek Borges, Ana Paula Santiago De Falco

A B C

AD: Adipic Acid. Figure 3. Photomicrographs of the cross sections of the obtained membranes: (A) flat membrane, temperature of the polymer solution = 23°C; (B) flat membrane, temperature of the polymer solution = 60°C; (C) hollow fiber: temperature of the polymer solution = 23°C.

A B

AD: Adipic Acid. Figure 4. Photomicrographs of the cross sections of the flat membranes obtained using additives: AD (5% wt) and addition: (A) distilled water (1% wt); (B) activated carbon (1% wt).

REFERENCES

KANG, G. D.; CAO, Y. M. Application and modiﬁcation of ZHANG, Y.; WANG, R. Gas liquid membrane contactors for acid gas poly(vinylidene ﬂuoride) (PVDF) membranes, a review. Journal of removal: recent advances and future challenges. Current Opinion Membrane Science , v. 463, p. 145-165, 2014. https://doi.org/10.1016/j. in Chemical Engineering, v. 2, n. 2, p. 255-262, 2013. https://doi. memsci.2014.03.055 org/10.1016/j.coche.2013.03.008

| 21 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 18-21 MATERIALS DEVELOPMENT

PRODUCTION OF GLYCEROL ETHERS ADDITIVES FOR IMPROVING COLD FLOW PROPERTIES Produção de aditivos à base de éteres do glicerol para melhoria de propriedades de fluxo a frio

Paula Araujo de Oliveira 1, Claudio José de Araujo Mota 2, Bianca Peres Pinto 3

Abstract: #e search for renewable energy sources became one of Resumo: A busca por fontes energéticas renováveis tornou-se um the primary themes of the century. In 2004, the National Biodiesel dos temas primaciais do século. Em 2004, criou-se o Programa Production and Use Program ( Programa Nacional de Produção e Uso Nacional de Produção e Uso do Biodiesel para viabilizar a produção do Biodiesel – PNPB) was created to enable the production and use e o uso do biodiesel no país. Na produção do biodiesel é gerada cerca of biodiesel in the country. In biodiesel production, about 10% of de 10% de glicerina como coproduto. Investigam-se novas rotas tec- glycerin is generated as a co-product. New technological routes are nológicas para promover o aproveitamento e a comercialização desse being investigated to promote the use and commercialization of this poliálcool excedente nos setores industriais. Uma maneira atraente é surplus polyalcohol in the industrial sectors. An attractive way to a síntese de éteres do glicerol empregando diferentes álcoois. As rea- deal with this excess is the synthesis of glycerol ethers combined ções são realizadas em série (combinações possíveis dos álcoois em with di%erent alcohols. #e reactions are performed in series (possi- pares) em reator Parr (batelada), utilizando resina ácida e alterando ble combinations of alcohols in pairs) in a Parr reactor (batch), using as variáveis operacionais. Posteriormente, os produtos são alquilados acid resin and changing the operating variables. Subsequently, the parcialmente com sulfato de metila. Os éteres do glicerol são anali- products are partially alkylated with methyl sulfate. Glycerol ethers sados por cromatogra!a gasosa acoplada à espectrometria de massas are analyzed by gas chromatography coupled with mass spectrome- e testados como aditivos oxigenados visando melhorar as proprieda- try and tested as oxygenated additives to improve the cold ]ow pro- des de ]uxo a frio do biodiesel (palma e sebo), que possui ponto de perties of biodiesel (palm and tallow), which has a freezing point of congelamento de aproximadamente 18°C, causando problemas de approximately 18°C, causing ]ow problems during winter, mainly in ]uxo no inverno, sobretudo no sul do país. the southern regions of the country. Palavras-chave: Glicerina 1. Eteri!cação 2. Propriedades de ]uxo Keywords: Glycerin 1. Etheri!cation 2. Cold Flow Properties 3. a frio 3.

1. Master in Chemical and Biochemical Process Engineering from the School of Chemistry, Universidade Federal do Rio de Janeiro. Assistant of the materials Division of the Naval Engineering Directorate – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 2. PhD in Organic Chemistry, Universidade Federal do Rio de Janeiro. Director and Professor at the Institute of Chemistry and Engineering of Chemical and Biochemical Processes, Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 3. PhD in Sciences by the Institute of Chemistry, Universidade Federal do Rio de Janeiro. Researcher at the Laboratory of Reactivity of Hydrocarbons, Biomass and Catalysis – Rio de Janeiro, RJ – Brazil. E-mail: [email protected]

| 22 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 22-25 Paula Araujo de Oliveira, Claudio José de Araujo Mota, Bianca Peres Pinto

1. INTRODUCTION of biodiesel. #e proposal consists of synthesizing glycerol ethers with di%erent alcohols (methanol, ethanol, tert-bu- Currently, the high depletion of natural resources that has tanol, and n-butanol) as reagents. #e multiple reactions in been occurring for years shows that it is essential to look for series (in pairs of two alcohols) occur in a Parr reactor, in a renewable energy sources. According to Resolution No. 45/2014 batch regime, in the presence of a solid heterogeneous acid of the National Agency of Petroleum, Natural Gas and Biofuels, catalyst (Amberlyst-15 resin), by varying the nature and biodiesel is de!ned as a “fuel produced from the transesteri!ca- order of the reagent employed. #e reactions are carried out tion and/or esteri!cation of fatty substances, of fats of vegetable by means of the possible combinations of the alcohols in two or animal origin, composed of alkyl esters of long carboxylic reactions. #e reaction time stipulated for the present work acids chain” (ANP, 2014). Glycerin is obtained as a by-product was four hours for each reaction. Subsequently, the products of the biodiesel formation reaction. Biodiesel can be used pure obtained will be partially alkylated (remaining hydroxyl) with or mixed with mineral diesel for combustion in engines, since the addition of methyl sulfate. #e products of the reactions its physical-chemical speci!cation is similar to that of diesel of both processes aim at good conversion and selectivity of fossil origin (PARENTE, 2003). Currently, the mandatory and are analyzed basically by gas chromatography coupled proportion of biodiesel in diesel is 12% (B12) (ANP, 2020). #e with mass spectrometry — a technique of great importance, main barrier presented for its use is the large supply of glycerin since it combines the resolving power of gas chromatography generated. Approximately 10% of the total volume of biodiesel with the high selectivity and sensitivity of the mass detector. produced corresponds to glycerol. As a result, new processes Subsequently, the glycerol ethers obtained through the pro- are emerging to take advantage of surplus glycerin that aim to posed reactions will be tested as oxygenated additives in order transform it into higher value-added products (essential for the to improve the cold ]ow properties of biodiesel. Pour point, economic sustainability of the biodiesel industry), since its o%er cold !lter clogging, and mist tests will be performed. is increasingly abundant in the market due to the growing gov- ernment incentive for the production of biodiesel. A promising recovery for this trihydroxylated alcohol is its transformation 3. RESULTS AND DISCUSSION into ethers. #e process produces compounds with less viscosity and polarity and, consequently, with greater volatility compared #e results obtained in Table 1 were analyzed and it was to pure glycerin (SILVA et al., 2009). #e addition of glycerol opted for conditions in which the glycerol conversion was ethers has the purpose of improving diesel cetanage, since the moder-ately high (approximately 60–80%), although it did use of additives causes better fuel performance, facilitating the not reach values close to 100%. It was expected that, after cold start of the engine (NOUREDDINI, 2001). Numerous the end of the !rst reaction, its medium would still contain investigations to obtain glycerol ethers with alcohols and few enough glycerol and glycerol monoether to react with the studies with methyl sulfate have been reported worldwide. second alcohol of the subsequent reaction. #e objective #e research interest was due to the fact that both are excel- was to be able to attribute to only one molecule the bene!ts lent etherifying agents for glycerol and low-cost raw materials, of using three di%erent reagents (two di%erent alcohols and thus optimizing the process for the production of oxygenated methyl sulfate) and to form products with great potential to additives for fuels. It is possible to form up to !ve isomers of be used as oxygenated additives. ether: two monosubstituted ethers, two disubstituted ethers, Tert-butanol has signi!cantly higher reactivity compared and one tri-substituted one. to the other investigated alcohols (carbocation stabilized by three methyl groups). #us, tert-butylation is successful at lower reaction temperatures (KLEPACOVA, 2005). #e reaction 2. MATERIALS AND METHODS temperature of 120°C was admitted when tert-butanol was used as a reagent, and 180°C for the other alcohols. #e glyc- #e present work focuses mainly on developing additives erol/alcohol molar ratio used was 1:3, as this excess amount based on glycerol ethers to improve the cold ]ow properties of reagent has already shown satisfactory results in terms of

| 23 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 22-25 Paula Araujo de Oliveira, Claudio José de Araujo Mota, Bianca Peres Pinto conversion (approximately 60–80%) and selectivity to the of the present study, since it will have fewer OH groups free glycerol monoether, establishing that the subsequent reaction to be partially alkylated with methyl sulfate (the next step, also contributes positively for the newly formed molecule. third reaction in series, will be harmed). Regarding selectiv- #e series reaction, using a new reagent (alcohol) in the ity, the reagent that showed the best selectivity for glycerol products obtained previously, aims to increase the propor- diethers was tert-butanol (carbocation stabilized more eas- tion of glycerol diethers and, thus, alkylate hydroxyls of the ily by the three methyl groups), as there was an increase in remaining glycerol and the monoether formed in the !rst the amount of glycerol diethers formed (from 20 to 40%) reaction. #e intention is to investigate the contribution of to the detriment of the monoethers in reaction 1. Since the adding the second reagent to the conversion and selectivity molecule is bulky, there is more di$culty in the formation of the studied process. It was observed that, in all cases, the of triethers. Changing the order of the reagents in the serial addition of the second alcohol contributed to the e$ciency of reactions did not show divergent results. the process (signi!cant increase in conversion — around 20%). #e best results obtained in terms of conversion were the reaction that used methanol in the !rst reaction and ethanol as 4. CONCLUSIONS a reagent in the second reaction (Figure 1), since the attack on the remaining hydroxyls from reaction 1 and the accessi- Until now, the reactions of etheri!cation of glycerol with bility to acidic sites become easier, as both reagents are small di%erent alcohols (methanol, ethanol, tert-bu-tanol and n-bu- compared to the others used and there is no signi!cant steric tanol) that enable the formation of mono-, di- and triether impediment. However, a higher formation of glycerol triethers products in this work are derived from glycerol. Amberlyst 15 was found, which is not interesting for the future proposal resin presented excellent performances and was the catalyst

Table 1. Results obtained through the first etherification reaction with several alcohols: tested operational variables (base case). Molar Selectivity (%) Alcohol Conversion (%) ratio Mono Di Tri By-products Methanol 30 96 4 - - Ethanol 18 100 - - - 1:3 Terc-butanol 81 72 25 3 - n-butanol 6 74 22 - 4 T = 120°C Methanol 40 93 7 - - Ethanol 20 78 17 5 - 1:6 Terc-butanol 88 70 27 3 - n-butanol 17 80 15 - 5 Methanol 59 74 23 3 - Ethanol 85 80 20 - - 1:3 Terc-butanol 92 65 20 5 n-butanol 72 80 15 - 5 T = 180°C Methanol 76 74 16 10 - Ethanol 95 80 12 8 - 1:6 Terc-butanol 98 70 20 10 - n-butanol 85 71 24 - 5

| 24 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 22-25 Paula Araujo de Oliveira, Claudio José de Araujo Mota, Bianca Peres Pinto

Serial reactions – Methanol (reagent 1)

100 90 80 70 60

% 50 40 30 20 10 0 methanol (reaction 1) ethanol (reaction 2) terc-butanol (reaction 2) n-butanol (reaction 2)

conversion mono di tri by-products

Figure 1. Results obtained using methanol as a reagent in reaction 1.

used in the study. Serial reactions were also studied, chang- (with the selected operational conditions). #us, the trans- ing the nature and order of alcohols by means of possible formation of glycerol into products with higher added value combinations of the reactants in two reactions. #e second promotes the biodiesel production chain and provides the reaction proved to be e$cient for the present study, obtain- industry with several relevant compounds originating from ing greater conversions and selectivities to the glycerol ethers a renewable reagent.

REFERENCES

AGÊNCIA NACIONAL DO PETRÓLEO, GÁS NATURAL E PARENTE, E.E. Biodiesel: uma aventura tecnológica nem país BIOCOMBUSTÍVEIS (ANP). Resolução n. 45, de 25 de agosto de 2014. engraçado. Fortaleza: Tecbio, 2003. Dispõe sobre a especiﬁcação do biodiesel contida no Regulamento NOUREDDINI. System and process for producing biodiesel fuel with Técnico ANP nº 3 de 2014 e as obrigações quanto ao controle da reduced viscosity and a cloud point below thirty-two (32) degrees qualidade a serem atendidas pelos diversos agentes econômicos que Fahrenheit. USPTO Patent Full . Patent n. 6,174,501, p. 4-14, 2001. comercializam o produto em todo o território nacional. Rio de Janeiro: ANG, 2014. Disponível em: http://legislacao.anp.gov.br/?path=legislacao- SILVA, C.R.B.; GONÇALVES, V.L.C.; LACHTER, E.R.; MOTA, C.J.A. anp/resol-anp/2014/agosto&item=ranp-45-2014. Acesso em: mar. 2020. Etheriﬁcation of Glycerol with Benzyl Alcohol Catalyzed by Solid Acids. Journal of the Brazilian Chemical Society , v. 20, n. 2, p. 201-204, KLEPACOVA, A.; MRAVEC, D.; HAJEKOVA, E.; BAJUS, M. Tert- 2009. https://doi.org/10.1590/S0103-50532009000200002 butylation of glycerol catalysed by ion-exchange resins. Applied Catalysis A , v. 294, n. 2, p. 141-147, 2005. https://doi.org/10.1016/j. apcata.2005.06.027

| 25 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 22-25 MATERIALS DEVELOPMENT

SOL-GEL COATING CONTAINING ENCAPSULATED CORROSION INHIBITOR FOR PROTECTION OF AA2024 ALLOYS Revestimento sol-gel contendo inibidor de corrosão encapsulado para proteção de ligas AA2024

Lais de Campos Barbieri 1, Brenda Diane Pereira Martins 2, Renata Batista Rivero Garcia 3, Elizabete Yoshie Kawachi 4

Abstract: AA2024-T3 alloys are widely used in the aerospace Resumo: As ligas AA2024-T3 são amplamente utilizadas na industry due to their low density, good fracture resistance and high indústria aeroespacial devido a sua baixa densidade, boa resistência mechanical strength. However, the presence of copper makes these a fraturas e alta resistência mecânica. No entanto, a presença de alloys prone to corrosion. Several coatings that protect against and cobre torna essas ligas propensas à corrosão. Diversos revestimentos inhibit corrosion have been studied to replace traditional Cr 6+ - protetores e inibidores contra corrosão vêm sendo estudados para based treatments, and promising results are being found with sel- substituir os tradicionais tratamentos a base de Cr 6+ , e resultados f-healing coatings. #is work studied the encapsulation of cerium promissores estão sendo encontrados com revestimentos self- nitrate in hollow urea-formaldehyde polymer microparticles and healing. Este trabalho estudou o encapsulamento de nitrato de their incorporation in a silicon-based sol-gel system for subse- cério em micropartículas ocas de polímero de uréia-formaldeído quent !lm deposition on AA2024 substrates, with the intention e incorporação dessas em sistema sol-gel à base de silício, para of evaluating the self-repairing / inhibiting properties of cerium in posterior deposição na forma de !lme sobre substratos de AA2024, the protection of AA2024 alloy against corrosion. com intenção de avaliar as propriedades autorreparadoras/ Keywords: Corrosion. Self-healing. AA2024. inibidoras do cério na proteção da liga AA2024 contra corrosão. Palavras-chave: Corrosão. Self-healing . AA2024.

1. Master’s student in Materials Chemistry at the Aeronautics Institute of Technology – São José dos Campos, SP – Brazil. E-mail: [email protected] 2. Master’s student in Materials Chemistry at the Aeronautics Institute of Technology – São José dos Campos, SP – Brazil. E-mail: [email protected] 3. PhD in Aeronautical Engineering from Aeronautics Institute of Technology. Physics Teaching Officer at the Air Force Academy – Pirassununga, SP – Brazil. E-mail: [email protected] 4. PhD in Chemistry from Universidade Estadual de Campinas. Associate professor at the Aeronautics Institute of Technology – São José dos Campos, SP – Brazil. E-mail: [email protected]

| 26 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 26-30 Lais de Campos Barbieri, Brenda Diane Pereira Martins, Renata Batista Rivero Garcia, Elizabete Yoshie Kawachi

1. INTRODUCTION 2. MATERIALS AND METHODS

Corrosion is a problem that can start in a crack and #e synthesis of microparticles with encapsulated cerium compromise the properties of the material, leading to nitrate was adapted from Matsuda et al. (2016). In a beaker catastrophic failures. For example, the crash of an air- 14.25 g of Span 80, 0.7 4g of Tween 80 and 200 g of glycerin craft that occurred in 1994 in Brazil, in which half of its were mixed. In another beaker, 4 g of urea-formaldehyde resin crew died. The accident reports indicated that the pro- previously prepared, 4 g of methyl 3-mercaptopropianate, 10 g peller blade had broken due to pitting and intergranular of propanone, 2 g of deionized water, 5 g of hydrogen perox- corrosion, which facilitated the propagation of fatigue ide and 1 g of Ce(NO 3)3 were added. #us, the oil phase was cracks, in addition to compromising the properties prepared in the !rst beaker, and the water phase was prepared of the affected part (LI; CALLE, 2007). Corrosion of in the second beaker. #e water phase was added to the oil metals causes substantial financial losses, and requires phase, stirred mechanically for 30 min. #e prepared emulsion efforts to limit their impact. It is estimated that about was homogenized with an ultrasonic probe for 10 min and 2 to 5% of the national gross domestic product (GDP) mechanically stirred again for 3 h at 50°C. Afterwards, the of countries like the United States is spent on dam- emulsion was vacuum !ltered to separate the microparticles, age directly caused by corrosion (KOCH et al., 2006; which were washed with hexane. SAMADZADEH et al., 2010). To prepare the sol-gel !lm, a glass reactor with a water Aluminum alloys, especially AA2024 (Al-Cu-Mg-Mn), cooling system was used, in which 15.10 mL of tetraethox- are commonly used in the aerospace industry due to their ysilane (TEOS), 15.90 mL of 3- (trimethoxysilyl)propryl properties, such as low density, good fracture resistance methacrylate (TMSPM) and 10 mL of isopropanol were and high mechanical resistance. The intermetallic par- added. #e mix was left under ultrasonic agitation with ticles that are deposited in the alloy’s grain contours are amplitude ( λ) of 50% and frequency (f) of 0.5 Hz, for responsible for these properties, but also decrease resis- 5 min. After that, 3 mL of nitric acid and 3.4 ml of deion- tance to the natural corrosion of aluminum. Coating sys- ized water were added, and ultrasonic stirring was contin- tems to protect these alloys used to employ chromium ued for 30 min, at a temperature of 10°C. as the best corrosion inhibiting agent, but the Cr 6+ ions To incorporate the microcapsules, 0.1 g of the microcap- present in these treatments have carcinogenic and pol- sules were added with cerium nitrate (III) encapsulated in luting properties, thus this use has been banned, accord- 5 mL of isopropanol. #is solution was left in an ultrasound ing to regulations of the Restriction on Hazardous bath for 15 min to deglomerate the capsules. With the help Substances (RoHS – 2002/95/EC), of the European of a micropipette, 3 mL of the solution was collected with Union, and of the Toxic Substances Control Act (TSCA) the suspended microparticles, and the sample collected was and Occupational Safety and Health Administration added to the sol. #e application of the coating with the (OSHA), of the United States (EUROPEAN UNION, microparticles incorporated in the AA2024 plates was done 2003; BARAL; ENGELKEN, 2002). Thus, other types by means of a dip-coater. #e alloy was immersed in the solu- of protective and self-repairing coatings have been used tion while still in the form of a sol, with an immersion/emer- to control and prevent damage caused by corrosive pro- sion speed of 1 mm/s and permanence of 1 s in the solution. cesses. A promising system is the incorporation and con- #e metal plate wetted with the !lm and the microparticles trolled release of a corrosion inhibitor in coatings, which was placed to dry in order to evaporate the solvents and to can protect the material before irreversible damage occurs allow the gel !lm to form. (BUHROW et al., 2011). Cerium has been shown to be To characterize the morphology of the coatings and iden- an efficient inhibitor against corrosive processes, being tify the presence of cerium (III) microparticles, the Tescan a promising substitute for chromium, since it does not Vegan 3 scanning electron microscope (SEM) was used. have the polluting and carcinogenic properties of chro- The linear polarization test was performed using an mium VI (ZAID et al., 2008). Autolab ® PGSTAT30 potentiostat in a three-electrode

| 27 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 26-30 Lais de Campos Barbieri, Brenda Diane Pereira Martins, Renata Batista Rivero Garcia, Elizabete Yoshie Kawachi electrolytic cell: the reference electrode (saturated Ag/AgCl), homogeneous and does not present considerable cracks, the working electrode (uncoated substrate and substrate coated as shown in Figure 1A, which indicates that the micro- with and without microparticles) and the counter electrode capsules have been incorporated and that, although (platinum spiral). #e exposed area tested was 0.39 cm 2, and they are not homogeneously distributed, they are not the electrolyte used was a NaCl solution at 0.1 mol/L, nat- agglomerated (Figure 1B). In Figures 1C and 1D, the urally aerated at room temperature. sol-gel coating is observed without incorporation of the microcapsules, which presents some irregularities, but apparently no cracks and defects that expose the 3. RESULTS AND DISCUSSION substrate surface. Figure 2 shows the linear polarization curves of sam- Figure 1 shows the micrographs obtained by SEM ple AA2024 without any coating (2A), sample AA2024 for the 2024 aluminum substrates coated with silicon with the sol-gel !lm without the presence of microparti- sol-gel. The coating containing the microcapsules is cles (2B), and sample AA2024 with the sol-gel !lm and

A B

C D

Figure 1. Scanning electron microscope (SEM) micrographs of AA2024 substrates with sol-gel coating, (A and B) with the addition of microparticles and (C and D) without the addition of microparticles.

| 28 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 26-30 Lais de Campos Barbieri, Brenda Diane Pereira Martins, Renata Batista Rivero Garcia, Elizabete Yoshie Kawachi with embedded microparticles (2C). Table 1 shows the the degradation of the sample tends to be faster than values of potential and current density obtained based on that of the AA2024 alloy coated with sol-gel only, and the graphical analysis. tends to be slower than that of the alloy AA2024 with- It is possible to observe that the presence of the !lm out any coating. without the microparticles left the AA2024 alloy less susceptible to corrosion, as the potential increased from -0.54 V (AA2024 without coating) to -0.30 V (with sol- Table 1. Electrochemical parameters of the AA2024 gel coating). When the microparticles were incorporated, substrate uncoated and coated with different sol- gel systems. the potential became positive, presenting a signi!cant Potential Current density increase, which means, from the thermodynamic point of Substrate (V) (A/cm 2) view, an increase in the resistance of the metallic part to -8 corrosive attacks. AA2024 uncoated -0,54 5,75x10 The current density value of the film system with AA2024 coated -0,30 1,81x10-10 the addition of microcapsules is lower than in pure with sol-gel alloy samples, but higher than in the alloy with film AA2024 coated with sol-gel without the addition of microcapsules. Kinetically, +0,16 2,83x10-8 once the corrosive process has started in the AA2024 containing microcapsules sample coated with sol-gel-containing microcapsules,

1E-8 (A) (B)

1E-4 ) 2 ) 2 1E-9

1E-6 1E-10

Log J (mA/cm Log 1E-11 Log J (mA/cm Log 1E-8

1E-12 -0,65 -0,60 -0,55 -0,50 –0,45 -0,40 -0,40 -0,35 -0,30 -0,25 -0,20 -0,15 Potential E vs. Ag/AgCl (V) Potential E vs. Ag/AgCl (V)

1E-6 (C)

1E-7 ) 2

1E-8

1E-9

Log J (mA/cm Log 1E-10

1E-11 0,00 0,05 0,10 0,15 0,20 0,25 0,30 Potential E vs. Ag/AgCl (V)

Figure 2. Polarization curves of the AA2024 substrate (A) uncoated, (B) coated with sol-gel without microcapsules and (C) coated with sol-gel with the addition of microcapsules.

| 29 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 26-30 Lais de Campos Barbieri, Brenda Diane Pereira Martins, Renata Batista Rivero Garcia, Elizabete Yoshie Kawachi

4. CONCLUSION by corrosion will be slower than the degradation presented by the uncoated AA2024 alloy. #e sol-gel coating with incorporated microparticles caused a considerable increase in the potential value, which indicates less susceptibility of the alloy to corrosive processes. 5. ACKNOWLEDGEMENTS #erefore, the results are in line with the expectations for the protection of the AA2024 alloy against corrosion. From a This work was carried out with the support of the kinetic point of view, the !lm without microparticles proved Coordination for the Improvement of Higher Education to be more e$cient, since its current density value is lower. Personnel Brazil (CAPES) — Process nº 88882.161950/2014- However, the current density value of the AA2024/!lm/ 01 — and the National Council for Scienti!c and Technological microparticle system indicates that the degradation caused Development (CNPq) — Process no. 311215/2016-5.

REFERENCES

BARAL, A.; ENGELKEN, R.D. Chromium-based regulations and LI, W.; CALLE, L.M. Microencapsulation for corrosion detection and greening in metal ﬁnishing industries in the USA. Environmental control. NASA Technical Report , 2007. Science and Policy , v. 5, n. 2, p. 121-133, 2002. https://doi.org/10.1016/ MATSUDA, T.; JADHAV, N.; KASHI, K.B.; JENSEN, M.; SURYAWAHSHI, S1462-9011(02)00028-X A.; GELLING, V.V. Self-healing ability and particle size eect of BUHROW, J.; LI, W.; JOLLEY, S.; CALLE, L.M. Microencapsulation encapsulated cerium nitrate into pH sensitive microcapsules. Progress technology for corrosion mitigation by smart coatings. NASA in Organic Coatings , v. 90, p. 425-430, 2016. https://doi.org/10.1016/j. technical report , 2011 porgcoat.2015.10.021

EUROPEAN UNION. Directive 2002/95/EC of the European SAMADZADEH, M.; HATAMI BOURA, S.; PEIKARI, M.; KASIRIHA, Parliament and of the council on the restriction of the use of certain S.M.; ASHRAFI, A. A review on self-healing coatings based on micro/ hazardous substances in electrical and electronic equipment. OFF. J. nanocapsules. Progress in Organic Coatings , v. 68, n. 3, p. 159-164, Eur. Union, 2003. 2010. https://doi.org/10.1016/j.porgcoat.2010.01.006

KOCH, G.H.; BRONGERS, M.P.; THOMPSON, N.G.; VIRMANI, Y.P.; ZAID, B.; SAIDI, D.; BENZAID, A.; HADJI, S. Eects of pH and PAYER, J.H. Corrosion costs and prevention strategies in the United chloride concentration on pitting corrosion of AA 6061 aluminium States, FHWA-RD-01-156, U.S . Washington, D.C.: Department of alloy. Corrosion Science , v. 50, n. 7, p. 1841-1847, 2008. https://doi. Transportation, Federal Highway Administration, 2006. org/10.1016/j.corsci.2008.03.006

| 30 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 26-30 MATERIALS DEVELOPMENT

BUILDING HIGH CAPACITY CYBER ARTIFACTS USING HIGH PRIVILEGED RESOURCES FROM MICROPROCESSORS Construção de artefatos cibernéticos de alta capacidade utilizando os recursos de alto privilégio de microprocessadores

William Augusto Rodrigues de Souza 1

Abstract: #is work discusses how to use the System Management Resumo: Este trabalho discute como usar o modo de gerenciamento Mode (SMM) on Intel processors to establish a set of require- do sistema (SMM) dos processadores Intel a !m de estabelecer um ments and from those requirements proposes a generic architec- conjunto de requerimentos e, a partir dele, propor uma arquitetura ture that can be used to build high capacity security tools or cyber genérica que pode ser usada para construir ferramentas de segurança weapons, which can take full control of the host computational ou armas cibernéticas, as quais teriam o controle total do sistema com- system. A proof-of-concept was built and successful experiments putacional onde fossem instaladas. Uma prova de conceito foi cons- were conducted showing the feasibility of this proposal and the truída e os experimentos realizados foram bem-sucedidos, demons- validity of the requirements. trando a viabilidade da proposta e a validade dos requerimentos. Keywords: Cyber weapons. Security tools. System management Palavras-chave: Armas cibernéticas. Ferramentas de segurança. mode. Intel processors. Modo de gerenciamento do sistema. Processadores Intel.

1. INTRODUCTION over the other tasks running on the system. In addition, they need to have overall vision or reach of the system and ability to handle the very environment in which it Computer security requires a process of contin- is operating, increasing the chances of success in accom- uous improvement, including the ability to perform plishing its tasks. verification tasks on a target system to minimize the A common security task that demands isolation and risk of potential threats becoming successful attacks. high privilege is to measure the integrity of system com- An example of this type of verification is the use of ponents, such as essential operating system and virtual antivirus software. machine monitor !les, operating in virtualized environ- In this context, some concepts are fundamental, such ments such as data centers and cloud computing struc- as: isolation, privilege and vision. Security tools should tures. #e security check can be done statically (when operate in isolation to minimize the risk of the same the target item is only stored in memory, without being attacks they are trying to avoid. They must have high updated or running) or dynamic (during the execution of privilege to take action at the necessary time, with priority the target item).

1. Lieutenant Commander (T) of the Naval Systems Analysis Center - Rio de Janeiro, RJ - Brazil. E-mail: [email protected]

| 31 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 31-35 William Augusto Rodrigues de Souza

In general, to measure integrity in a static way, a secu- architectures. It was not by accident that SMM was released rity tool can execute a cryptographic service to generate on the Intel 386SL, a mobile version of Intel 386 (INTEL, a hash code of the target. To verify integrity, it is neces- 2001; 2003; 2015). Mobile computers need a robust way sary and sufficient to generate a new target hash code a for system power management. #us, SMM was designed moment ahead in time and compare the two codes. If the to provide power management and other system manage- codes are the same, no change have occurred and integrity ment functions, such as hardware control. Prior to SMM, is preserved. Otherwise, it could be said that there was most of these functions were performed by in-circuit emu- breach of integrity, though not necessarily malicious. It is lation (ICE) tools. up to the system administrator to make this judgment, A key feature of SMM is its high privilege; SMM’s since, in theory, they have control over non-malicious executive software, called SMI handler, has higher priv- changes in the environment. Performing dynamic mea- ileges than the operating system or virtual machine surement is much more complicated because the target monitors. It is installed by the Basic Input and Output will, of course, vary its content over time, generating dif- System (BIOS) into an isolated and protected memory ferent hash codes, which does not necessarily indicate a area, allowing SMI handler manages critical system tasks breach of integrity. In both cases, the security tool must with the highest priority. This is the focus of this work, be installed in an isolated environment so as not to be given that, if we override the SMI handler with a secu- attacked and violated, to have sufficient privileges and rity tool or a cyber weapon, these artifacts will, in theory, good view (reach) of the target items distributed around have the same features as the SMI handler. This paper the computing system. discusses how to accomplish this substitution, establish- #us, the problem that security analysts are facing is the ing the necessary requirements and proposing the archi- deployment of a security tool in an isolated computer sys- tecture that the artifact (security or cyber weapon) should tem environment with high privilege and good view of the use to be successful. targets in order to minimize the risk of attacks. In this work, it is possible, without loss of generality, to The resources available in the computer system when replace the terms “security” or “security tool” by “malicious the processor starts operating in the system manage- code” or “cyber weapon” and thus host a powerful malicious ment mode (SMM) provide appropriate environment code in the SMM memory that would have unrestricted and and characteristics to be used by a security tool, such protected access to the entire computer system. For practi- as: isolation, high privilege and good vision of the sys- cal purposes, the term “security tool” is used more often than tem. In this article, only Intel processors are considered; “malicious code” (or “cyber weapon”) in this work. however many of the features valid for Intel proces- In recent years, SMM has been successfully exploited sors are valid for processors from other manufacturers, in different ways, for example: by avoiding SMM pro- such as AMD. tection mechanisms using the cache poisoning tech- Because of its resources — isolated memory area, high nique (DUFLOT; ETIEMBLE; GRUMELARD, 2006; privilege and execution priority, including over the oper- WOJTCZUK; RUTKOWSKA, 2009b) or exploiting ating system, in addition to full system-wide view and a failure of implementation of a specific SMI handler, access — SMM has been used for purposes other than those leading to a breach in the mechanisms of Intel Trusted described by Intel (INTEL, 2014a; 2014b; 2014c), contrib- Execution Technology (Intel TXT), paving the way for uting to the arms race between attackers and defenders of the successful exploitation of the Xen virtual machine computer systems. monitor (one of the most well-known and used in the #e SMM was introduced in the IA-32 architecture, market) (WOJTCZUK; RUTKOWSKA, 2009a). On the in the Intel 386SL processor, launched in 1990 (INTEL, other hand, security researchers have presented new 2001; 2003; 2014c; 2015), to provide resources to man- tools that take advantage of the capabilities of SMM, age the processor’s host computer system. Since then, such as: HyperSentry (AZAB et al., 2010), used to mea- SMM has been a standard item in the IA-32 and IA-64 sure the integrity of virtual machine monitors; Strongly

| 32 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 31-35 William Augusto Rodrigues de Souza

Isolated Computing Environment (SICE) (AZAB; NING; probes, through programs in the languages C and assembly, ZHANG, 2011), to provide an isolated execution envi- to several computers equipped with di%erent models of Intel ronment; and AppCheck (WANG; SUN; STAVROU, processors to identify the details exposed in the manuals of 2012), to protect applications by inspecting the respec- these products. tive codes stored in physical memory. A set of requirements and a threat model were established, and a generic architecture was designed to be used in the construction of artifacts. A proof of concept was con- 2. OBJECTIVE structed using the C language.

Despite the considerable number of works on SMM resources, there are no studies investigating SMM itself and 4. WORK RESULTS its foundations. #e works found report the use of resources and ways to take advantage of some characteristics of the #e requirements established to construct artifacts using SMM, but there is a lack of studies dedicated to under- the resources of the SMM were the following: standing this system, establishing requirements for its use • Requirement 1 – Small. #ere are 32512 bytes available in security tasks and de!ning an architecture for its use as a for the SMI handler code and its data in the isolated and platform for security tools. protected SMM memory, called SMRAM. SMRAM In particular, the security tools based on the SMM fea- can be up to 4 Gbytes in size, but it is recommended tures studied in this research do not take advantage of all to keep the minimum size so as not to a%ect the trans- these attributes, failing notably in exploiting their isolation parency of the tools. If you look at security tools that and transparency. In addition, many of these tools do not use SMM resources, most of them use a similar archi- respect SMM constraints and limitations, such as the maxi- tecture, taking into account a remote machine to ana- mum latency time (the time the operating system “standstill”, lyze collected data and other management functions. waiting for the SMI handler to run). In general, the architec- #is architecture solves the issue of limited space by ture of these tools is modular, and only one of the modules modularizing the tools. Although it alleviates a func- is placed in the isolated and protected memory area of the tional problem, this creates a security problem because SMM, communicating with other modules stored in unpro- it opens new attack locations, such as the communi- tected areas of the system, which lack the isolation o%ered by cation channel, drivers and devices used to allow com- the SMM and can be attacked, as well as other system items. munication with the remote machine. In addition, such #us, this research investigates SMM in the context of tools lose the SMM isolation protection when using Intel processors to establish a set of requirements and, from other modules outside the SMRAM. #us, an SMM- it, propose a generic architecture that can be used to build based security application must be small enough to !t security tools or high capacity cybernetic weapons, which the smallest size available. would have total control of the computer system where they • Requirement 2 – Fast. Intel speci!es that maximum were installed. Finally, a proof of concept is built to measure latency, i.e. the maximum time the computing system the integrity of a Xen virtual machine monitor !le. Such can remain running in SMM mode, should be less measurement is limited to the minimum necessary to prove than 150 microseconds to minimize the risk of system the concept of architecture. executive software time-outs, among other problems (INTEL, 2014d). #us, the execution time for each call to execute the artifact must be less than or equal 3. WORK METHODOLOGY to 150 microseconds. • Requirement 3 – Persistent. SMRAM is volatile, #e methodology used in this work consisted of research and rebooting the system will wipe all its contents. in the scienti!c literature and in the manuals of Intel and in Thus, code and SMM-related data need to be reloaded

| 33 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 31-35 William Augusto Rodrigues de Souza

at each boot process. An artifact that uses the fea- generated. A common approach to triggering an SMI tures of SMM, therefore, needs to be embedded in and starting such software is to write in the input and the BIOS (or equivalent entity) since the BIOS con- output port “0xB2H”. As this is a well-known approach, tains the SMM-related code and its data and loads an attack can be used to prevent such an action by them into the SMRAM during the computer system identifying the code signature and then denying the boot process. use of that path. Thus, an SMM-based artifact must • Requirement 4 – Cooperative. #e functions of the take advantage of any SMI generated to begin per- SMI handler need to be preserved as they have import- forming its work. On the other hand, whenever the ant tasks to perform. Any SMM-based software must tool needs to start, it should be able to start an SMI preserve the original functions of the SMI handler by in different ways. adding its own code to the code of that executive soft- • Requirement 8 - Complete. As discussed in require- ware without replacing any part of it. Upon entering ments 1 (small, so that all parts are within SMRAM) SMM, the processor looks for the !rst instruction to and 5 (isolated, all parts within SMRAM and thus be executed at the address indicated by the SMBASE + without remote machine use), some security tools need 0x8000 register (by 0x38000 by default) in the SMRAM, a remote machine to analyze collected data and other where the SMI handler is located, which implies that management functions. Such tools must retain part any SMM-based artifact must be a modi!ed version of of their code in the operating system or in the virtual the SMI handler. machine monitor. HyperSentry (AZAB et al., 2010), • Requirement 5 – Isolated. SMI handler — and there- for example, uses an agent deployed in the code base fore any SMM-based software — performs its tasks of the virtual machine monitor. While solving some without notifying or being recognized by the operat- problems, these implementations lose the bene!ts of ing system (GRAWROCK, 2008). In addition, since using SMM, such as isolation and transparency. #us, the operating system is stopped for as long as the pro- a SMM-based software must have all the functionality cessor is in the SMM, running the SMM-based soft- to perform its tasks and all necessary data completely ware is transparent to any other software on the sys- installed in SMRAM. tem. #erefore, producing an artifact to use SMM and placing it in the memory area outside of SMRAM is, A proof of concept was built using the C language to at least, counterintuitive, since the main motivation for measure the integrity of an essential !le (xend-con!g.sxp) using SMM is to bene!t from its features, such as iso- of the Xen virtual machine monitor. #e experiments were lation and transparency successfully performed on 13 machines, with details of the • Requirement 6 – Resistant. Cache poisoning is an e%ec- processors and chipset of each machine (Table 1). tive attack on SMM, made possible by manipulating Memory Type Range Registers (MTRRs) to make the SMRAM “cacheable” (meaning that SMRAM data can 5. CONCLUSION be then copied into the cache memory). To prevent such an attack, the SMRR Interface engine, provided #is work presented a research on the SMM in Intel pro- by Intel, must be used to protect the related MTRR cessors in order to establish a set of requirements and, from registers. #us, any resource available in the system to it, propose a generic architecture that can be used to con- enhance SMM security must be employed. It is warned struct security tools or high capacity cyberweapons, which that not all Intel processors have support for the SMRR would have total control of the computing system where Interface. they were installed. • Requirement 7 – Be called for any interruption of A proof of concept was built and the experiments per- system management. To start any SMM-based soft- formed were successful, demonstrating the feasibility of the ware, a system management interrupt (SMI) must be proposal and the validity of the requirements.

| 34 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 31-35 William Augusto Rodrigues de Souza

Table 1. Machines used in the experiments. OEM Processor MCH ICH

Gigabyte Dual Core E2160 945P ICH7

Gigabyte Pentium 4 945P ICH7

Asus Tek Pentium 4 Not identiﬁed ICH7

Dell i5-2500 Sandy Bridge H67

Asus Atom N270 82945GSE ICH7-M

Foxconn Atom N270 E7500 ICH3-M

HP i5-650 Clarkdale P55

Fujitsu Pentium 4-M 548 82845 ICH3-M

HP i7-2670QM Sandy Bridge HM65

Acer i5-3450 Ivy Bridge B75

Compaq Pentium III-M 82830M ICH3-M

REFERENCES

AZAB, A.M.; NING, P.; WANG, Z.; JIANG, X.; ZHANG, X.; SKALSKY, ______. Intel 64 and IA-32 Architectures Software Developer’s Manual . N.C. Hypersentry: Enabling stealthy in-context measurement of Denver: Intel Corporation, 2014a. v. 1. hypervisor integrity. In : ACM CONFERENCE ON COMPUTER AND ______. Intel 64 and IA-32 Architectures Software Developer’s Manual . COMMUNICATIONS SECURITY, 17., 2010, Nova York. Anais… Nova Denver: Intel Corporation, 2014b. v. 2. York: ACM, 2010. p. 38-49. ______. Intel 64 and IA-32 Architectures Software Developer’s Manual . AZAB, A.M.; NING, P.; ZHANG, X. SICE: A hardware-level strongly Denver: Intel Corporation, 2014c. v. 3. isolated computing environment for x86 multi-core platforms. In : ACM CONFERENCE ON COMPUTER AND COMMUNICATIONS SECURITY, ______. The Intel BIOS Implementation Test Suite (BITS), version 18., 2011, Nova York. Anais… Nova York: ACM, 2011. p. 375-388. http:// 1090. Denver: Intel Corporation, 2014d. doi.org/10.1145/2046707.2046752 ______. Intel 64 and IA-32 Architectures Software Developer’s Manual . DUFLOT, L.; ETIEMBLE, D.; GRUMELARD. Using CPU system Denver: Intel Corporation, 2015. v. 3. management mode to circumvent operating system security WANG, J.; SUN, K.; STAVROU, A. Hardware-assisted application functions. In : CANSECWEST SECURITY CONFERENCE, 7., 2006. integrity monitor. In : HAWAII INTERNATIONAL CONFERENCE ON Anais… 2006. SYSTEM SCIENCE, 45., 2012. Anais… HICSS, IEEE Computer Society, GRAWROCK, D. Dynamics of a trusted platform: a building block 2012. p. 5375-5383. http://doi.org/10.1109/HICSS.2012.299 approach. 2ª ed. Hillsboro, OR: Intel Press, 2008. WOJTCZUK, R.; RUTKOWSKA, J. Attacking intel trusted execution INTEL. Intel IA-32 Architecture Software Developer’s Manual . Denver: technology. In : BLACK HAT CONFERENCE., 2009. Anais … Las Intel Corporation, 2001. v. 3. Vegas, 2009a.

______. Intel IA-32 Architecture Software Developer’s Manual . Denver: WOJTCZUK, R.; RUTKOWSKA, J. Attacking SMM memory via intel Intel Corporation, 2003. v. 3. CPU cache poisoning. 2009b.

| 35 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 31-35 OPERATIONAL RESEARCH

A PANORAMA OF THE USE OF OPERATIONAL RESEARCH IN THE SUBMARINE WAR ENVIRONMENT Um panorama do uso de pesquisa operacional no ambiente de guerra submarino

Leonardo Antonio Monteiro Pessôa 1, Edilson Fernandes de Arruda 2, Laura Bahiense 3

Abstract: #is paper presents a panorama of the applied use Resumo: Este artigo apresenta um panorama do uso de metodo- of Operations Research methodology on submarine warfare. logias de pesquisa operacional aplicadas à guerra em ambiente sub- A bibliographic review regarding the Scopus database is presen- marino. Apresenta-se revisão bibliográ!ca referente à base Scopus, ted, exploring the most used techniques, clustering the documents no tocante às técnicas mais utilizadas, apresentando os países de and presenting the countries of origin for a better understanding. origem e agrupando os documentos para melhor entendimento. #e paper provides a deeper knowledge about how OR has been O trabalho contribui para conhecimento mais profundo sobre as used in this context, presenting the most recent developments and maneiras nas quais as ferramentas de PO estão sendo utilizadas no tendencies, and may guide future Brazilian research on this theme. domínio da guerra submarina apresentando as tendências e desen- Keywords: Operations research. Defense. Military. ASW. volvimentos mais recentes, podendo servir como orientação para futuras pesquisas brasileira nesse tema. Palavra-chave: Pesquisa operacional. Defesa. Militar. Guerra A/S.

1. PhD in Production Engineering by the Alberto Luiz Coimbra Institute of Postgraduate Studies and Engineering Research at Universidade Federal do Rio de Janeiro. Project Manager at the Naval Systems Analysis Center – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 2. PhD in Electrical Engineering by Universidade Estadual de Campinas. Adjunct Professor of the Production Engineering Program at the Alberto Luiz Coimbra Institute of Postgraduate Studies and Engineering at Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 3. PhD in Systems and Computer Engineering by the Alberto Luiz Coimbra Institute of Postgraduate Studies and Engineering Research at Universidade Federal do Rio de Janeiro. Associate Professor of the Systems and Computer Engineering Program and the Transport Engineering Program at Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected]

| 36 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense

1. INTRODUCTION the study. However, the Scopus base is considered as a relevant repository for the Brazilian scienti!c community, as #e present work aims to construct an overview of the it includes periodicals of great relevance, allowing to verify use of operational research (OR) methods applied to war in the most important methodological advances and to identify the submarine environment. For this purpose, a bibliographic trends in temporality, origin and areas of knowledge involved, research using the Scopus platform is presented. #e choice in addition to contributing to the identi!cation of applica- of this bibliographic base is justi!ed by its wide scope, as tion possibilities and methodological gaps. well as its acceptance in the Brazilian scienti!c community. For this purpose, the methodology used for the bib- It should be noted that this is still a preliminary research, liographic survey is detailed below. In the initial phase, the despite the significant number of references found. It is method used to conduct the bibliographic review consisted intended, therefore, to enrich the research by consulting of the following steps: other known bibliographic bases, such as Web of Science, • Scopus base search using the terms “submarine” “warfare” ScienceDirect etc. and “Operations Research” or “Operational Research”; In addition, it is intended to search bibliographic docu- • Presentation of results in relation to: temporality; source; ments (articles from journals, conferences etc.) indexed in bib- areas of knowledge; liometric databases with information regarding the country of • Mapping of the most relevant keywords using the origin. We also want to infer the temporal behavior of publica- VOSviewer software; tions related to the topic, the most frequent types of publica- • Mapping of the most relevant words in the abstract using tion, in addition to the most explored methodological aspects. the VOSviewer software; In order to methodologically characterize the structure • Classi!cation of items according to the domain and the of this article, the second section presents a brief theoreti- OR tool used. cal foundation, detailing the approach employed in carrying out the present work. Section 3 presents the results, focusing on the identi!ed references. Finally, section 4 brings a brief 3. RESULTS discussion of the results and the conclusion. Table 1 shows the entry used for consultation, which initially resulted in 23 articles. 2. METHODOLOGY Figure 1 shows the number of publications surveyed according to the year, covering the period from 1984 to 2018. #e interest in the anti-submarine war environment is #ere is an irregularity in the publications, with several years related to the beginnings of the organized OR, speci!cally to lacking representation. the developments of Blackett’s “circus” (TREFETHEN, 1970). However, it is noted that there is a propensity to increase On the other hand, we see that this environment is sig- publications from 2016 onwards. From that date on, there ni!cant for the Brazilian context, when observing the impor- are publications in all years. In addition, it should be noted tance given by the National Defense Strategy (BRASIL, that the largest number of publications occurred in 2018. 2012) to the submarine war environment. #ere are refer- #is data suggests an increase in interest in the subject in ences in this context (PESSÔA, 2016), mainly theses and recent years and a possible upward trajectory of the number dissertations, but which are not on the basis considered for of publications on the subject.

Table 1. Search terms and articles found. Base Query Found Docs TITLE-ABS-KEY ( ( ”submarine” AND ”warfare”) SCOPUS 23 AND ( ”Operations Research” OR ”Operational Research”) )

| 37 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense

Figure 2 places the articles according to the area of knowledge. #ere is a predominance of articles related to the areas 10% 7% of engineering and decision science, which represent more 7% than 60% of the documents found. Figure 3, in turn, shows a predominance of literature 5% 3% produced in the United States and China. Despite the indi- 27% cation of six entries with no de!ned origin, when analyzing 8% 2% the authors’ a$liation, which is presented in Figure 4, it is 2% observed that this predominance is coherent. Figure 5 presents the map of the occurrences of the cat- aloged keywords using the VOSviewer software. In order to 37% build it, it was necessary to delete the document (KARATAS; CRAPARO; AKMAN, 2018), since a corruption of the “.csv” !le, from Scopus, was detected when the document was kept. Engineering #e minimum limit of three occurrences was established Decision Sciences for each keyword. With that, only 14 keywords remained, Business, Management and Accounting considered as the most important ones, of the total of 175. Computer Science Still in Figure 5, the terms “anti-submarine warfare (asw)” and “anti-submarine warfares” are noted. #ese terms appeared Mathematics separately, but when uni!ed, they would have a number of Social Sciences occurrences equal to that of the term “submarines” (eight Arts and Humanities occurrences). Chemical Engineering Figure 6 brings a di%erent perspective when representing Multidisciplinary the map of the most frequent associations of words found in the titles and abstracts of the researched articles. #e refer- Source: Scopus. ences to “system” and “probability” are striking, which often Figure 2. Documents per area.

Documents by year of publication 5

1 Number of documents

0 1991 2011 1987 1997 1992 1985 1993 1995 2017 1988 2012 1989 2015 2013 1998 1986 2018 1996 1999 1984 2016 1994 2014 1990 2010 2001 2007 2002 2003 2005 2008 2006 2009 2004 2000

Source: adapted from Scopus. Figure 1. Number of documents found per year of publication.

| 38 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense play a prominent role in the connection between the terms A3liation “asw”, “antisubmarine warfare”, and “target” and the terms 4 “convoy” and “path”. However, the analysis of this mapping is not enough to 3 show the methods of OR used. #erefore, we proceeded to analyze the abstracts in order to extract more information 2 about this, which are presented in Table 2. #e repetition of a document is veri!ed (LIU; XUE; WANG, 2013), so that 1 there are 22 di%erent ones. 0 #ere is a preponderance of the purposes related to: detection, combat engagement, and measurement of e%ectiveness. Even in the documents dealing with historical aspects, the analyses understood the improvement of the tactics through Corp Vitro the scienti!c method. Curtin University Curtin Wright-Patterson AFB Wright-Patterson ESAMS Aerospace Ltd Aerospace ESAMS Wright State University State Wright

4. DISCUSSION AND CONCLUSION Research Naval of O3ce Naval Postgraduate School Postgraduate Naval PLA Dalian Naval Academy Naval Dalian PLA National Defense University Defense National

#e literature found, despite being irregular in time, shows

renewed interest in recent years. #e war in the submarine Japan of Academy Defense National Defence Research Agency, Farnborough Agency, Research Defence

Origin Technology of Institute Air-Force States United 7 Space and Naval Warfare Systems Center San Diego San Center Systems Warfare Naval and Space Source: Scopus. 6 Figure 4. Documents per affiliation.

decision making 4 probability 3 military optimization operations 2 defence 1 studies helicopters submarines 0 warfare (asw) warfare

computer anti-submarine simulation China Japan

Turkey ordnance warships Australia

Undeﬁned operations Netherlands South Korea research United States United anti-submarine United Kingdom United warfares Source: Scopus. Figure 3. Documents per country. Figure 5. Keywords.

| 39 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense

environment is a development !eld centered in the United asw probability convoy States and China and in institutions linked to the Navy. anti-submarine #e concentration of documents in areas of knowl- warfare path target system edge focused on the methods of engineering, mathematics, and decision-making are noticed, which are directed to direct application in the anti-submarine war, which Figure 6. Most frequent connections. aims to !nd, chase, detain, and damage or destroy enemy

Table 2. References found. Methodological Source Problem Purpose elements mentioned integral geometry - geometric Karatas et al.(2018) probability - Monte Carlo use of sonobuoys submarine detection Simulation nonzero-sum two-player network interdiction allocation of patrol resources in opposition to Hew and Yiap(2018) game -Strong Stackelberg chokepoints the submarine equilibrium. eocient ways of operating a Yoash et al.(2018) \textit{search pattern} helicopter equipped with sonar submarine detection dipping Kim et al.(2018) simulation sequential experimentation combat engagement examining the impact of Duney(2017) historical comparison measures and countermeasures historical lessons on anti-submarine warfare Hohzaki(2016) \textit{search theory} \textit{search game} literature review enectiveness anti-submarine warfare operation measurement - Liu et al.(2013) \textit{information entropy} by surface ship information quality assessment calculation of torpedoes’ orientation angle in relation Xia et al.(2012) mathematical model combat engagement to the target - acoustic search torpedoes measuring the Glassborow et al. enectiveness of ISR discrete optimization \textit{shortest path problem} (2005) systems - convoy protection supervision of high-demand naval units during low intensity discussion / Koch (2005) - operations, development of defense news technologies for conducting anti- submarine warfare \textit {two-sided, multi-stage Vermeulene Van \textit {search theory} search problem} search for submarine detection DenBrink (2005) enemy submarine by naval force Continue...

| 40 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense

Table 2. Continuation. Methodological Source Problem Purpose elements mentioned discussion on weapons of mass discussion / Schneider Jr.(2004) - destruction defense news launching of the first F 310 Fridtjof Nansen-class Aegis- discussion / Lok (2004) - equipped frigate for the Royal defense news Norwegian Navy (RNoN) Comprehensive Maritime Patrol operational Brennan and mathematical model Aircraft Systems Simulation enectiveness Denton(2004) (COMPASS) (prediction) Champagne et al. \textit{search theory, historical case study involving submarine detection - (2003) agent-based simulation} onensive search engagement in combat oceanography, supporting tactical decision detection - Incze (1996) acoustics and OR making in military missions engagement in combat development of Neveu and Pignon simulation method and tools for \ simulation underwater warfare (1993) textit{design} and development systems for surface ships new approach to quantify the Crawford and probability of successful anti- enectiveness probability Sawyer(1991) submarine warfare mission measurement without destroying the submarine presenting technical and scientific advice to the Supreme Allied Commander Atlantic (SACLANT) discussion / Martin (1991) - and North Atlantic Treaty defense news Organization (NATO) nations in the anti-submarine warfare submarine torpedo attack, Urban(1990) - specifically salvo engagements combat engagement against multiple targets representing tactical situations Knapp et al.(1987) computer simulation involving two opposing combat engagement submarines maximizing the probability of detection of a Eagle (1984) Markov process detecting a submarine submarine

submarines. The topic of detection and search of sub- It is recognized that there is also a signi!cant portion marines is seen as the most robust branch of discussion. not directed to the OR itself, in which there are news and #ere are also highlighted aspects of combat engagement discussions about defense. #ese contributions serve as an and e%ectiveness measurement. element for contextualizing the political-military interest in In this universe of submarine warfare, reasonable diversity the theme, helping to construct the panorama. can be noted in the methods employed, which can be related #e existence of publications related to the theme in to optimization and heuristic models, mathematical analyzes, important OR vehicles demonstrates the interest of the theme simulations, and theory of games. #is diversity shows that for literature and, given the approach of the Navy to federal the problems encountered can be confronted under di%er- teaching and research institutions, shows that it is interesting ent approaches. to develop national research on this speci!c theme.

| 41 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 Leonardo Antonio Monteiro Pessôa, Edilson Fernandes de Arruda, Laura Bahiense

REFERENCES

BRASIL. Estratégia Nacional de Defesa . Brasil, 2012. KNAPP, B.M.; DUDLEY, A.R.; RYDER, J.S. Modelling techniques for simulation of submarine engagements. Journal of the Operational BRENNAN, J.F.M.; DENTON, A.L. Compass—the veriﬁcation and Research Society, v. 38, n. 10, p. 891-898, 1987. https://doi.org/10.1057/ validation of an operational analysis model for use in the prediction jors.1987.154 of Nimrod MRA4 operational eectiveness. Journal of the Operational Research Society, v. 55, n. 4, p. 413-421, 2004. https://doi.org/10.1057/ KOCH, A. US Navy outlines expeditionary command plans. Jane’s palgrave.jors.2601663 Defence Weekly, 2005.

CHAMPAGNE, L.; CARL, E.G.; HILL, R.R. Search theory, agent- LIU, J.; XUE, C.-Y.; WANG, Y.-J. A method of information quality based simulation, and U-boats in the bay of Biscay. In : Simulation assessment of ship ASW operation. In : INTERNATIONAL SYMPOSIUM Conference, 2003. Anais [...]. 2003. v. 1. p. 991-998. https://dx.doi. ON OPERATIONS RESEARCH AND ITS APPLICATIONS IN org/10.1109/WSC.2003.1261521 ENGINEERING, TECHNOLOGY AND MANAGEMENT, 11., 2013. Anais […]. China: Institution of Engineering and Technology, 2013. p. 1-4. CRAWFORD, J.D.; SAWYER, F.L. ASW measures of eectiveness. Vitro Technical Journal , Norfolk, v. 9, n. 1, p. 12-19, 1991. LOK, J.J. Norway’s Fridtjof Nansen-class frigates to provide a key strategic advantage. Jane’s International Defence Review, 2004. DUFFEY, R. B. Submarine warfare and intelligence in the Atlantic and Paciﬁc in the Second World War: comparisons and lessons MARTIN, R.L. SACLANTCEN: submarine threat neutralizer. Sea learned for two opponents. Journal for Maritime Research, v. 19, n. 2, Technology , Londres, v. 32, n. 5, 1991. p. 143-167, jul. 2017. NEVEU, D.; PIGNON, J.-P. Simulation for underwater warfare systems EAGLE, J.N. Optimal search for a moving target when the search path design. Revue Technique, v. 25, n. 2, p. 669-682, 1993. is constrained. Operations Research , v. 32, n. 5, p. 1107-1115, 1984. PESSÔA, L.A.M.; TEIXEIRA, L.S.; GUEDES, M.J.M.; MARTINS, E.R.; SOUZA, GLASSBOROW, D.; CACCETTA, L.; DAVIS, C.; REHBOCK, V. Estimating A.J.N. Pesquisa Operacional na Marinha Do Brasil: O Casnav, seu presente the value of intelligence, surveillance and reconnaissance in Manoeuvre e perspectivas. In : SIMPÓSIO BRASILEIRO DE PESQUISA OPERACIONAL, Warfare. In : INTERNATIONAL CONGRESS ON MODELLING AND 48., 2016, Vitória. Anais […]. SBPO, 2016. p. 2302-2312, 2016. SIMULATION: ADVANCES AND APPLICATIONS FOR MANAGEMENT AND DECISION MAKING, 2005. Anais […]. Rockingham, 2005. SCHNEIDER JR., W. A 21st-century role for nuclear weapons. Issues in p. 1758-1764. Science and Technology, v. 20, n. 3, p. 55-60, 2004.

HEW, P.; YIAP, N. Optimally randomized patrolling of chokepoints for TREFETHEN, F.N. A history of Operations Research. In : GARRETT, theatre antisubmarine warfare. Military Operations Research , v. 23, R.A.; LONDON, J.P. (orgs.). Fundamentals of Naval Operations n. 1, p. 49-56, 2018. Analysis. Annapolis: United States Naval Institute, 1970. p. 188-205.

HOHZAKI, R. Search games: Literature and survey. Journal of the URBAN, C.D. Design and evaluation of a tactical decision aid. In : IEEE Operations Research Society of Japan , v. 59, n. 1, p. 1-34, 2016. https:// INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS, doi.org/10.15807/jorsj.59.1 1990. San Diego: US Naval Ocean Syst Center, 1990. p. 812-814.

INCZE, B.I. Tactical oceanographic battlespace. Sea Technology , VERMEULEN, J.F.J.; VAN DEN BRINK, M. The search for an alerted Arlington, v. 37, n. 8, p. 43-49, 1996. moving target. Journal of the Operational Research Society , v. 56, n. 5, p. 514-525, 2005. https://doi.org/10.1057/palgrave.jors.2601847 KARATAS, M.; CRAPARO, E.; AKMAN, G. Bistatic sonobuoy deployment strategies for detecting stationary and mobile underwater targets. XIA, Z.-J.; ZHANG, X.-H.; XU, L.-Z. Study on orientation angle of Naval Research Logistics , v. 65, n. 4, p. 331-346, 2018. https://doi. two acoustic homing torpedoes relative to target with parallel salvo org/10.1002/nav.21807 launched by submarine. Binggong Xuebao/Acta Armamentarii , Liaoning, v. 33, n. 5, p. 636-640, 2012. KIM, J.H.; SEO, K.-M.; LEE, T.-E.; CHOI, B.W. Achieving new insights into combat engagement analysis via simulation-based sequential YOASH, R.B.; ATKINSON, M.P.; KRESS, M. Where to dip? Search experimentation. Military Operations Research , v. 23, n. 4, p. 51-80, pattern for an antisubmarine helicopter using a dipping sensor. 2018. https://doi.org/10.5711/1082598323451 Military Operations Research , v. 23, n. 2, p. 19-39, 2018.

| 42 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 36-42 DECISION-MAKING PROCESS

A DECISION-MAKING SYSTEM FOR THE MARINE CORPS BASED ON MULTI-CRITERIA DECISION SUPPORT METHODS Uma sistemática de decisão para o Corpo de Fuzileiros Navais baseada em métodos de apoio multicritério à decisão

Leonardo Amorim do Amaral 1, Newton José Ferro 2, José Calixto dos Santos Junior 3, Sérgio Douglas Oliveira Bentes 4, Newton José Ferro Junior 5

Abstract: #is study deals with the application of a Resumo: O presente trabalho trata-se da aplicação de uma aborda- decision-making support approach based on the convergence gem sobre o apoio à decisão com base na convergência dos méto- of the ELECTRE III and AHP methods, in the selection of dos Eliminação e Escolha como Expressão da Realidade na sua ver- an operational environment, dubbed a platform, that can best são III (Election et Choix Traduisant La Realite — ELECTRE III) meet the expeditionary needs of the Brazilian Marine Corps. e de Análise Hierárquica do Processo (Analytic Hierarchy Process #e purpose was not to compare the two methods, but to help — AHP), na seleção de um meio operativo, denominado de plata- the decision-making process by taking advantage of the speci- forma, que melhor possa atender às necessidades expedicionárias do !c characteristics of each method and to identify possible con- Corpo de Fuzileiros Navais (CFN). Não teve como propósito compa- vergences. In order to make the use of the ELECTRE III and rar os dois métodos, mas sim auxiliar o processo de decisão no apro- AHP methods feasible as an aid to the decision-making process veitamento das características especí!cas de cada método e identi!- regarding the same problem through the search of possible con- car as possíveis convergências. A !m de tornar exequível a utilização vergences, the authors empirically elaborated a table for assig- dos métodos ELECTRE III e AHP como auxílio à decisão de um ning the judgments for insertion in the AHP method, based on mesmo problema, por meio da busca de possíveis convergências, os the di%erence of the medians of the weights attributed by the autores elaboraram, empiricamente, uma tabela de atribuição dos jul- ten judging specialists of the established criteria, which could gamentos para inserção no método AHP, baseado na diferença das evolve to another parametric approach, as a process of conti- medianas dos pesos atribuídos pelos dez especialistas julgadores dos nuous improvement. critérios estabelecidos, os quais poderão evoluir para outra abordagem Keywords: Multi-Criteria Decision Analysis. Election et Choix paramétrica, como um processo de melhoria contínua. Traduisant La Realite. Analytic Hierarchy Process. Decision Palavra-chave: Multicritério de Apoio à Decisão. Election et Support. Choix Traduisant La Realite. Análise hierárquica. Apoio à decisão.

1. Master’s degree in Nuclear Engineering from the Alberto Luiz Coimbra Institute for Graduate Studies and Engineering Research at Universidade Federal do Rio de Janeiro. Science, Technology and Innovation Assistant at the Marines Material Command – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 2. PhD in Nuclear Engineering from the Alberto Luiz Coimbra Institute for Graduate Studies and Engineering Research at Universidade Federal do Rio de Janeiro. Science, Technology and Innovation Advisor to the Marine Material Command – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 3. Master in Naval Sciences at Escola de Guerra Naval. Manager of 4 × 4 Light Armored Vehicles of the Marine Material Command – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 4. Assault Amphibian Unit Leader Course at Assault Amphibian School Battalion Training Command. Manager Assistant of 4 × 4 Light Armored Vehicles of the Marine Material Command – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 5. Graduate in Engineering Research from the Alberto Luiz Coimbra Institute for Graduate Studies and Engineering Research at Universidade Federal do Rio de Janeiro. Master’s degree from the Alberto Luiz Coimbra Institute for Graduate Studies and Engineering Research at Universidade Federal do Rio de Janeiro – Rio de Janeiro, RJ – Brazil. E-mail: [email protected]

| 43 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

1. INTRODUCTION AMD methods aim to be an aid tool for analysts and decision makers in situations where there is a need to iden- In the !eld of studies on the decision-making process, sev- tify priorities based on multiple criteria that involve two or eral disciplines and methodologies have been developed to assist more alternatives (GOMES, 1999). in the construction of decision models, such as decision trees, Among these types of problems, it is common to classify game theory, linear programming and multicriteria analysis. the methods developed by the American School and French #e latter has a great integration with the Fuzzy Sets #eory; School. #e methods developed are based on prevalence rela- Fuzzy Logic; Arti!cial Neural Networks and Rough Sets tionships, and among them, the family methods stand out #eory, and its main characteristic is to scienti!cally treat the ELECTRE (ROY, 1968; 1978; ROY; SKALKA, 1984) and subjectivity inherent to the decision process (COSTA, 2005). PROMETHEE (BRANS; MARESCHAL; VINCKE, 1984). In this context, the Brazilian Marine Corps started a #e American School methods already have characteristics study to search for an operational means that provides troop that aim to reduce the various criteria through synthesis - shipments to complete speci!c missions that meet the min- in the vast majority of times, by means of a weighted sum. imum requirements needed for employment in the Marine Among these, we can highlight the AHP methods (SAATY, Operative Groups (GptOpFuzNav). Due to it being a strategic 1980), TODIM (GOMES, 1987) and MACBETH (BANA acquisition for military use, this article will not mention what E COSTA; VANSNICK, 1995; 1997). types of operational means were analyzed, their models and manufacturers, which will now be called simply as platforms. 2.1. ELECTRE METHOD #is study is seen as a classic complex decision problem, #e ELECTRE method was designed by Bernard Roy in in which several criteria may be necessary for a !nal choice response to de!ciencies in existing decision-making methods. between di%erent alternatives. #erefore, it is considered as ELECTRE has evolved through several versions, all based a problem of Multicriteria Decision Support (MDS), since, on the same fundamental concepts, but operationally slightly according to Almeida (2013), it consists of a situation where di%erent (ROY, 1991). there are at least two action alternatives to choose from, and ELECTRE is a well-known method that has a history of this choice is driven by the desire to meet multiple objectives, successful applications in various types of decision-making often con]icting with each other. situations. Its approach requires an input of criteria evalua- #us, this study aimed to present the methodology used in tions for the alternatives called Decision Matrix, informa- the process of choosing that platform, using MDS concepts, tion on preference, expressed as weights, thresholds and other particularly the ELECTRE III method and the AHP method, parameters (SEVKLI, 2010). developed for use by a single decision maker, adapted by the All ELECTRE-type methods involve two main procedures: authors for an approximation with multiple decision makers. the modeling of preferences with outranking relations, followed by an exploitation procedure (KAYA; KAHRAMAN, 2011). According to de Almeida (2013), one of the main char- 2. LITERATURE REVIEW acteristics that distinguishes ELECTRE from many other multiple solution methods is that it is not compensatory. Several studies deal with the application of methods to #is means that good results in some criteria do not com- support organizations’ decisions. #e search for a clear meth- pensate for one or more bad results in others. Another char- odology that involves rationalizing the decision has been a acteristic of ELECTRE is that it allows incomparability, goal to be achieved. It is very common to !nd studies involv- that is, when there is no clear evidence in favor of some kind ing decisions for the selection of equipment or personnel. of preference or indi%erence between di%erent alternatives. According to Roy (1993), the decision support process is Because, considering a decision problem between alternatives de!ned as an activity in which a facilitator, using scienti!c a, b and c, for example, knowing that alternative a is better procedures, helps to obtain elements of answer to the questions than b and than c, it becomes irrelevant to analyze the pref- asked to the players involved in a decision-making process. erences between alternatives b and c. #ese two actions can

| 44 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior perfectly remain incomparable, without degenerating the deci- Consistency Ratio (CR) is equal or less than 0.1, represent- sion support procedure. #us, the basic idea in this family of ing a variation of up to 10%. #is is because even when the methods is to emphasize the analysis of dominance relations. parity judgments are based on the experience and knowledge According to Almeida (2013), the methods of the of specialists, inconsistencies can occur, mainly in the case of ELECTRE family are applied in two main phases: build- the existence of a high number of judgments. ing outranking relations, establishing a pair-wise compari- It should be noted that during the development of the son of alternatives and exploiting the relationship, applying AHP method, Saaty (1990) considered the experiments car- a procedure to solve the problem according to the speci!c ried out in the 1950s by the psychologist George Miller, who problem to be addressed. identi!ed that, usually, people properly process information ELECTRE methods consider weights as a measure of containing few facts, of the order of seven, with standard the importance of each criterion for the decision maker, and deviation of two units, being in harmony with the stability not as a marginal rate of substitution, since the evaluations of of the main autovalue for small disturbances when n is small. each alternative in the di%erent criteria do not come together In the case study presented in this article, the number in an overall evaluation. of criteria evaluated exceeded the psychological limit mentioned, by using 11 criteria, due to the operational needs of 2.2. AHP METHOD #e AHP method allows for subjective decision making. #e AHP is based on the Newtonian-Cartesian method, Table 1. Saaty fundamental scale. whose idea is to understand the complexity of the whole Value Definition Rationale through fragmentation at the lowest and most detailed levels Both activities Equal possible, establishing synthesized relationships (MARINS; 1 contribute equally importance SOUZA; BARROS, 2009). to the objective. According to Costa (2002), the method can be divided Experience and Moderate judgment slightly into three phases: construction of hierarchies, de!nition of 3 importance favor one activity priorities and logical consistency. #e construction of hier- over another. archies allows overall understanding and evaluation in the Experience and systemic view of the problem with all existing components Strong judgment strongly 5 and interactions. Structuring in hierarchical levels is per- importance favor one activity formed by a separation into, usually, three basic levels: the over another. !rst level is de!ned as the general purpose of the analysis, One activity is taking as an example the choice of a platform that provides highly favored over the other; troop shipment; the second hierarchical level includes the Very strong 7 its dominance established criteria; and the third level, the available alter- importance in importance is natives (MARINS; SOUZA; BARROS, 2009). demonstrated In the priority definition phase, joint judgments are in practice. made to synthesize the relationships between the criteria The evidence favors one activity over and alternatives. #e elements of a given hierarchy level are Extreme 9 the other with the judged by pairs in light of an element from the previous level importance highest degree (TREVIZANO, 2005), according to the Saaty Fundamental of certainty. Scale (1991), as shown in Table 1. 2, 4, 6, 8 When looking for Lastly, in the logical consistency phase, the coherence Reciprocal Intermediate a compromise between the relationships of the previous phases is veri!ed, of previous values condition between in which the AHP method proposes to calculate the consis- values two deﬁnitions. tency of the judgments — considered acceptable when the Source: Saaty (1991).

| 45 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior the analyzed platforms. #us, we used the AHP approach with to be analyzed as a problem of Multicriteria Decision Support, ratings (SAATY, 2008), which considers the assessment in whose values are shown and de!ned in Table 2. categories, based on levels of intensity assigned to each crite- • Ballistic Defense (Bld): Technology used on military rion related to each alternative, instead of the classic approach. platforms to provide personal protection against ballistic impacts, according to the STANAG 4569 NATO standard. #eir values vary between protection levels 1 to 4; 3. CASE STUDY • Garrison (Ga): For the purposes of this article, the term garrison includes the garrison of the platform itself, plus For this study, six operating platforms from di%erent man- the number of military personnel who can board it; ufacturers were selected, of which four fully and two partially • Life Expectancy (Exp): Time in years that the platform met a document prepared by the Brazilian Marine Corps, will be used in operational activities. For this article, all called High Level Requirements for Systems (RANS), which alternatives had a useful life cycle of 20 years; describes the technical characteristics necessary for the use • Gross Weight (GW): Load in kilograms (kg) of the sum of these platforms. Based on the data extracted from RANS, of the payload plus the weight of the platform; a conceptual map was elaborated establishing the most rel- • Representativeness (Rep): Number of countries that have evant criteria that in]uence the operational performance of the platform in operation; these platforms when ful!lling their missions. Figure 1 shows • Space between axles (SBA): Horizontal distance in milli- the outline of the conceptual map elaborated, in which the meters (mm) in the direction of the transverse axis of the weighting for each criterion was established. For example, platform between the centers of the front and rear wheels; when comparing the gross weight criterion of the six plat- • Power-Weight (PW): Ratio between the platform’s weight, forms, the best will be the one with the lowest absolute value. in kilograms (kg), and its engine’s maximum power in On the other hand, when comparing the autonomy criterion, horsepower (hp). #e unit is kg/hp; the highest absolute value will be privileged. • Height (H): Distance in meters (m) in the direction of the Based on the concept map, 12 criteria were identi!ed for vertical axis of the platform, between its most extreme parts; choosing the desired platform, noting that the life expectancy • Length (L): Horizontal distance in meters (m), in the criterion was removed from the process, since it had the same direction of the longitudinal axis of the platform, between value for all six platforms under study, resulting in 11 criteria its most extreme parts;

Ballistic Garrison Life Expectancy Gross Weight Defense

Lower

Higher Higher Higher Payload Representativeness

Higher Higher Best Platform Lower Lower Space Price Between Axles Lower Lower

Higher Higher

Autonomy Length Height Power-Weight

Figure 1. Conceptual map of the criteria that will define the best platform to be chosen.

| 46 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

• Autonomy (Aut): Maximum distance in kilometers (km) must use adequate armored protection against the envisaged that the platform can travel, under de!ned conditions, enemy threats, as well as carry a minimum of cargo, including without the need for refueling; the number of military personnel, their equipment and weap- • Price (Pr): Cost of acquisition of the platform in thou- ons. Decision on the use of the AHP method was also due sands of dollars; to its single synthesis criterion, allowing the compensation • Payload (PL): Total load, in kilograms (kg), that can pos- between criteria. Finally, an analysis of the results obtained sibly be transported inside the platform, including the by the two MDS methods was carried out as a way of pre- garrison, equipment, ammunition, armament and other senting the deciding authority with a representative choice. accessories needed for the performance of the activities. 3.2. USAGE OF THE ELECTRE III METHOD 3.1. METHODOLOGY ADOPTED Before evaluating the criteria, a verbal scale and its respective #is study uses the AMD ELECTRE III method in numerical values were elaborated, as shown in Table 3, whose conjunction with the AHP method to select an operating construction of hierarchies is considered a fundamental step in platform that best meets the needs of a GptOpFuzNav in the human reasoning process. #e elaboration of the table in ful!lling their tasks. During the selection of the platform, question was necessary, since the ELECTRE III method allows a desirable characteristic was presenting a balanced perfor- the use of verbal scales and di%erent numerical values, unlike the mance in all requirements — that is, it should not be evaluated AHP method that uses parity judgment between criteria, using poorly on any item, a fundamental parameter for resources the standardized hierarchy scale, as shown in Table 1, presented dispatched to the Brazilian Marine Corps, due to the lack previously. #us, the criteria were evaluated according to the of prior knowledge of the conditions of terrains involved in degree of importance established by the preference scale in Table future operations. In this context, the use of the ELECTRE 3 by 10 specialists (Esp) on the platforms under study, in which III method is justi!ed in view of it being uniquely non-com- each one assigned a value for each criterion, as shown in Table 4. pensatory among the requirements used. #e medians were then calculated for each criterion and, Although the balanced performance among all criteria is together with the technical characteristics of each platform desirable, it is also true that, as a platform for exclusive military (alternative), they were inserted in the ELECTRE III software use, some characteristics may admit compensation between for the simulation. For this study, no indi%erence or prefer- the criteria, as is the case with armored protection and pay- ence threshold was considered, since the AHP method was load, since, in order to ful!ll a speci!c mission, a platform used later to carry out the compensation between the criteria.

Table 2. Relationship of platforms with the values for each criterion. Platform 1 Platform 2 Platform 3 Platform 4 Platform 5 Platform 6 Ballistic Defense 3 4 3 2 2 3 Payload 2,109 2,000 2,200 1,043 1,500 2,200 Garrison 8 9 10 4 5 5 Price 550 550 550 300 550 1,300 Autonomy 550 700 1,000 466 500 700 Length 5.3 6.0 5.5 4.5 5.3 5.4 Height 2.3 2.3 2.1 1.8 2.1 2.0 Power-Weight 36.40 30.16 19.72 34.61 29.92 29.40 Space Between Axles 2,946 3,570 3,790 3,300 3,230 3,530 Gross Weight 8,845 12,000 10,900 5,488 8,100 8,500 Representativeness 10 3 9 31 6 3

| 47 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

Based on the data entered, the ELECTRE III software • P: Strong preference; presented the following preference relationship between the • R: Impossibility of comparison. alternatives, as shown in Figure 2. Based on the preference relationship, it can be noted that #e analysis of the !nal pre-order matrix and the decreas- platform 3 overranks all others. On the other and, it is observed ing priority graph in Figure 2 show the selection of Platform that there is no preference between platforms 1 and 6. It can 3 as the most suitable, since it was the most preferred over all also be noted that platforms 2 and 4 could not be compared other platforms. Also, as can be seen on Table 4, this platform regarding preference, however they rank below platforms 1 had preference even though it didn’t show the highest value and 6. Platform 5 is ranked last. #e !nal pre-order matrix, for the ballistic defense criterion. However, it is observed illustrated in Table 5, shows the degree of preference among that on average, this platform has good technical character- the alternatives, indicating the relationships as follows: istics for the seven highest scored criteria, as is the case with • I: Indi%erence; payload, autonomy and garrison. #is demonstrates that the • P’: Weak preference; ELECTRE III method considers, according to the characteristics of the platforms combined with the values attributed to the listed criteria by the experts, that platform 3 will be Table 3. Values of preferences. the one that best meet the needs for the ful!llment of its Verbal Scale Numerical missions with greater reliability and balance. Values Absolutely weak Importance 1 3.3. USAGE OF THE AHP METHOD Extremely weak Importance 2 As previously mentioned, the AHP method was used to Weak Importance 3 verify the hierarchy between the platforms under study and Very little Importance 4 to compare the result with that obtained in the ELECTRE III method, although this is not common practice. #is is Little Importance 5 because in order to be chosen, in addition to showing a good Moderate Importance 6 performance in all criteria, the platforms under study also Relevant Importance 7 needed to show characteristics in which less relevant criteria Strong Importance 8 can be compensated by others with higher relevance, such as Extreme Importance 9 the ballistic defense and payload criteria, which are essential Absolute Importance 10 requirements in the chosen platform to, respectively, oppose

Table 4. Attribution of values to the criteria by experts (Exp.). Exp. 1 Exp. 2 Exp. 3 Exp. 4 Exp. 5 Exp. 6 Exp. 7 Exp. 8 Exp. 9 Exp. 10 Ballistic Defense 10 10 9 9 10 8 10 10 10 10 Payload 7 10 9 7 5 7 10 10 8 8 Power × Weight 10 6 9 9 9 9 7 5 8 9 Autonomy 9 5 10 8 8 8 8 5 9 8 Price 7 10 10 6 10 5 7 6 8 7 Garrison 9 6 8 8 5 7 7 6 7 8 Gross Weight 8 3 5 10 5 6 5 7 9 10 Height 7 6 6 7 7 7 9 5 6 8 Space Between Axles 5 3 5 8 9 7 7 8 6 10 Length 8 6 4 6 5 7 7 8 6 7 Representativeness 6 3 7 7 3 7 8 6 10 6

| 48 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior the !repower threats showed by the opposing force and to of the expert opinions presented in Table 4, the 11 listed cri- transport all military personnel, their equipment and arma- teria were compared two by two, using the di%erence between ments to ful!ll the mission. the medians of the values assigned by the 10 specialists for For the use of the AHP method, a matrix similar to each criterion. For this purpose, Table 7 was created, which Table 2 was extracted, which can be identi!ed as a decision allows interpolation between these di%erences, in order to Matrix, in which the !rst column corresponds to the criteria enable the parity judgment between the criteria, according and the six following lines represent the platform alternatives. to the Saaty scale for the AHP method. Each cell corresponds to a platform alternative classi!ed in #us, when the di%erence in medians between two com- its respective criteria. pared criteria is equal to 1.0, for example, the degree of parity In order to carry out the even comparison between the judgment between them will correspond to the value of 3 on criteria, the values in the decision matrix were normalized in the fundamental Saaty scale, shown in Table 1. each criterion, as shown in Table 6, eliminating the creation However, for the comparative analysis between the alter- of judgment matrices among the alternatives, since all criteria natives, no group decision was used, because the intensity have levels of quantitative intensity. In order to enable the use levels of the criteria were assigned quantitatively. Table 8 shows the construction of the weighting matrix (judgments) according to the methodology of approach. When making parity comparisons between the criteria, it Platform 3 is noted that the highest judgment value was 7 (much more important) in the Saaty fundamental scale was obtained when the platform’s ballistic defense was compared with length and representativeness, meaning that the !rst has a much Platform 1 Platform 6 higher degree of importance than the other two. In addition, it appears that the ballistic defense, power-weight and payload are the most predominant criteria over others. When inserting the platforms’ characteristic data with Platform 2 Platform 4 their respective values of parity judgment, the priority vector of the criteria was obtained, according to Table 9. When multiplying the normalized decision matrix (Table 6) by the priority vector (Table 9), the results of hierarchical Platform 5 classi!cation between the platforms analyzed was found, as shown in Table 10. Source: adapted from software ELECTRE III. #us, it appears that with the application of the AHP Figure 2. Relationship of preference between method, Platform 2 was classi!ed as the best alternative for alternatives. acquisition. #is result was expected, as this platform has a

Table 5. Final preorder matrix. Platform 1 Platform 2 Platform 3 Platform 4 Platform 5 Platform 6 Platform 1 I P P’ P P I Platform 2 P’ I P’ R P P’ Platform 3 P P I P P P Platform 4 P’ R P’ I P P’ Platform 5 P’ P’ P’ P’ I P’ Platform 6 I P P’ P P I

| 49 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior very strong weight in relation to the others in the criteria of The CI is calculated using the following expression ballistic defense, power-weight and payload. In addition, it (Equation 2): is observed that platforms 4 and 5 were the worst alternatives, due to their low weight in terms of ballistic defense CI = (λmax – n) / (n – 1) (2) and payload. It is worth mentioning that although platforms 1, 3 and In which: 6 ranked, respectively, third, fourth and second, the percentage n = the number of criteria in the matrix. di%erence between them was very small, in the thousandths place, allowing a technical tie to be considered between these #erefore, the value found for the CI was (Equation 3): platforms in the second place. CI = (11.8161 – 11) / (11 – 1) (3) 3.3.1. Analysis of the consistency ratio CI = 0.08161 #e weighting matrix consistency ratio is found by the ratio between the consistency index (CI) and the random Table 7. Assignment of judgments for inclusion in the consistency index (RI). Hierarchical Analysis (AHP). To !nd the CI of the weighting matrix, the largest autovec- Difference between Judgement tor of the weighting matrix λmax, was calculated, consider- median weights value ing the sum of the product of the values of the priority vec- 0 to 0.25 1 tor of each criterion by the total sum of each column of the weighting matrix (Equation 1). 0.26 to 0.50 2 0.51 to 1.0 3 λmáx = (0.304358 × 2.83) + (0.199817 × 5.967) + 1.01 to 1.50 4 (0.120965 × 12.082) + (0.108054 × 14.082) + 1.51 to 2.25 5 (0.046274 × 22.5) + (0.046274 × 22.5) + 2.26 to 3.0 6 (0.046274 × 22.5) + (0.046274 × 22.5) + (1) 3.01 to 5.00 7 (0.027236 × 32) + (0.027236 × 32) + 5.01 to 7.0 8 (0.027236 × 32) >7.0 9 λmáx = 11.8161

Table 6. Normalized decision matrix. Platform 1 Platform 2 Platform 3 Platform 4 Platform 5 Platform 6 SUM Ballistic Defense 0.1765 0.2353 0.1765 0.1177 0.1177 0.1765 1 Payload 0.1908 0.1809 0.1991 0.0944 0.1357 0.1991 1 Garrison 0.1951 0.2195 0.2439 0.0977 0.1219 0.1219 1 Price 0.1447 0.1447 0.1447 0.0789 0.1447 0.3422 1 Autonomy 0.1404 0.1788 0.2554 0.1189 0.1277 0.1788 1 Length 0.1656 0.1875 0.1719 0.1406 0.1656 0.1688 1 Height 0.1825 0.1825 0.1667 0.1429 0.1667 0.1587 1 Power-Weight 0.2020 0.1674 0.1094 0.1921 0.1660 0.1631 1 Space Between Axles 0.1447 0.1753 0.1861 0.1620 0.1586 0.1733 1 Gross Weight 0.1643 0.2229 0.2025 0.1019 0.1505 0.1579 1 Representativeness 0.1613 0.0484 0.1452 0.5000 0.0967 0.0484 1

| 50 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

Saaty (1991) proposed a table with the random consis- Based on the results in Table 12, it can be seen that plat- tency indexes (RI) in matrices of orders 1 to 15 calculated forms 4 and 5 had a predominance rating below the others in a laboratory, as shown in Table 11. in both methods, and can be considered as alternatives to By building the relationship between the CI/RI, the con- be discarded, because, in addition to not ranking well using sistency ratio of the weighting matrix between the criteria is obtained. Table 9. Criteria priority vector. 1º Ballistic Defense 0.304358 CR = 0.08161/1.51 2º Power-Weight 0.199817 CR = 0.0540 3º Payload 0.120965 4º Autonomy 0.108054 #is result is quite acceptable, as it is lower than 0.1. 5º Price 0.046274 #e same reasoning was used to calculate the CR among 6º Garrison 0.046274 the alternatives for each criterion, staying within the accept- 7º Space Between Axles 0.046274 able limit. 8º Height 0.046274 9º Gross Weight 0.027236 3.4. ANALYSIS OF RESULTS 10º Representativeness 0.027236 #e results obtained by applying the two methods used 11º Length 0.027236 in this study were consolidated in Table 12, already considering the technical tie that occurred in the AHP method between platforms 1, 3 and 6, mentioned above, in which Table 10. Hierarchy between platforms. it can be observed that platform 3 was predominant with Overall Classification Platform the ELECTRE III method and ranked second in the AHP. Performance Platform 2, in turn, was predominant in AHP and ranked 1º Platform 2 0.1922 third in ELECTRE III. Platform 6 ranked second in both 2º Platform 6 0.1767 methods, as well as platform 1. As for platform 4, it obtained 3º Platform 1 0.1766 the fourth priority in ELECTRE III and the !fth in AHP, 4º Platform 3 0.1757 while platform 5 ranked sixth and had the last priority in 5º Platform 4 0.1410 both methods. 6º Platform 5 0.1378

Table 8. Weighting matrix. Bld PL Ga Pr Aut L H PW SBA GW Rep Bld 1 5 6 6 5 7 6 3 6 7 7 PL 0.2 1 3 3 3 4 3 0.333 3 4 4 Ga 0.167 0.333 1 1 0.333 2 1 0.2 1 2 2 Pr 0.167 0.333 1 1 0.333 2 1 0.2 1 2 2 Aut 0.2 1 3 3 1 4 3 0.333 3 4 4 L 0.143 0.25 0.5 0.5 0.25 1 0.5 0.167 0.5 1 1 H 0.167 0.333 1 1 0.333 2 1 0.2 1 2 2 PW 0.333 3 5 5 3 6 5 1 5 6 6 SBA 0.167 0.333 1 1 0.333 2 1 0.2 1 2 2 GW 0.143 0.25 0.5 0.5 0.25 1 0.5 0.167 0.5 1 1 Rep 0.143 0.25 0.5 0.5 0.25 1 0.5 0.167 0.5 1 1 Total 2.83 12.082 22.5 22.5 14.082 32 22.5 5.967 22.5 32 32

| 51 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

Table 11. Random consistency index. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49 1.51 1.48 1.56 1.57 Source: Saaty (1991).

Table 12. Comparative result of the ELECTRE III and Hierarchical Analysis (AHP) methods. Platform 1 Platform 2 Platform 3 Platform 4 Platform 5 Platform 6 1º - AHP ELECTRE - - - 2º ELECTRE/AHP - AHP - - ELECTRE/AHP 3º - ELECTRE - - - - 4º - - - ELECTRE - - 5º - - - AHP - - 6º - - - - ELECTRE/AHP -

both AMD methods, they only partially met some of the to avoid the selection of a platform that does not meet the requirements listed in the RANS. #us, platforms 1, 2, 3 and needs of the Brazilian Marine Corps. 6 remained to be analyzed. In this context, the approximation used through the medi- In this context, it is noted that platforms 2 and 3 had pre- ans of each criterion that served as a weight for the ELECTRE dominance in AHP and ELECTRE III, respectively. It can III method and used for equal weighting between the criteria be concluded then that platform 3 has a more balanced per- based on the AHP method, reached its purpose of assisting the formance among the 11 criteria weighted in the study, due to decision-making process, because, among the six platforms of the non-compensatory characteristic of the ELECTRE III interest selected initially for ful!lling the minimum character- method. #e results obtained for platform 2 in the AHP were istics established in the high level system requirements, it was expected, as it has very strong compensation in relation to the identi!ed that platforms 4 and 5 had the worst classi!cations ballistic defense criterion, which was the one with the great- among the others, in both methods, and were thus discarded. est weight obtained in the priority vector. Platforms 1 and 6, On the other hand, platform 3 obtained the preference in turn, cannot be ruled out, as they achieved convergence in with the ELECTRE III method, while platform 2 was the the second hierarchical position in both methods. best hierarchically ranked with the use of the AHP method. In addition, although platforms 1 and 6 did were not the preferred alternatives in any of the methods applied, they 4. FINAL CONSIDERATIONS ranked second in preference in both methods. It is worth mentioning that the values of the judgments Due to its expeditionary characteristics, the Brazilian established empirically by the authors, presented in Table 7, were Marine Corps need platforms that allow for the use of intended to allow the feasibility of using the ELECTRE III and GptOpFuzNav in di%erent operating scenarios. #us, before AHP methods, and can be replaced by other values developed in the !nal decision, the deciding authority needs to de!ne a parametric approach, as a process of continuous improvement. whether the need for a balanced solution, characteristic of Finally, this study did not aim to compare the two meth- the ELECTRE III method, or a compensatory decision ods, but to assist the decision-making process based on the hierarchized by the AHP method, prevails for alternatives use of the speci!c characteristics of each method and to iden- in which the two methods used did not converge, in order tify possible convergences.

| 52 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 Leonardo Amorim do Amaral, Newton José Ferro, José Calixto dos Santos Junior, Sérgio Douglas Oliveira Bentes, Newton José Ferro Junior

REFERENCES

BANA E COSTA, C.A.; VANSNICK, J.C. Thoughts on a theoretical ROY, B. Classement et choix en presence de points de vue multiple: framework for measuring attractiveness by categorical based La methode ELECTRE. R.I.R.O. , v. 2, n. 8, p. 57-75, 1968. https://doi. evaluation technique (MACBETH). In : CLÍMACO, J.C.N. (Org). org/10.1051/ro/196802V100571 Multicriteria Analysis. Berlim: Springer-Verlag, 1997. ROY, B. Decision science or decision-aid science? European Journal BANA E COSTA, C.A.; VANSNICK, J.C. Uma nova abordagem ao of Operational Research , v. 66, n. 2, p. 184-203, 1993. https://doi. problema da construção de uma função de valor cardinal: MACBETH. org/10.1016/0377-2217(93)90312-B Investigação Operacional , v. 15, p. 15-35, 1995. ROY, B. ELECTRE III: Un algorithme de classements fonde sur une BRANS, J.P.; MARESCHAL, B.; VINCKE, P.H. PROMETHEE: a new family represenation ﬂoue des preferences en presence de criteres multiples. of outranking methods in MCDM. In : BRANS, J.P. (org.). Operational Cahier du CERO, v. 20, n. 1, p. 3-24, 1978. Research . Amsterdã: IFORS’84, 1984. p. 477-490. ROY, B. The Outranking Approach and the Foundation of ELECTRE COSTA, H.G. Introdução ao método de análise hierárquica . Rio de Methods. Theory and Decision , v. 31, p. 49-73, 1991. https://doi. Janeiro, 2005. 104 p. Notas de aula. org/10.1007/BF00134132

COSTA, H.G. Introdução ao método de análise hierárquica: análise ROY, B.; SKALKA, J.M. ELECTRE IS: Aspects methologiques et guide multicritério no auxílio à decisão. Niterói: H.G.C., 2002. d’utilisation. LAMSADE, n. 30, 1984.

DE ALMEIDA, A.T. de. Processo de Decisão nas Organizações: SAATY, T.L. Decision making for leaders . Pittsburg: University of Construindo Modelos de Decisão Multicritério. São Paulo: Atlas, 2013. Pittsburg, 1991.

GOMES, E.G. Integração entre Sistemas de Informação Geográﬁca SAATY, T.L. Decision making with the analytic hierarchy process. e Métodos Multicritério no Apoio à Decisão Espacial . Dissertação International Journal of Services Sciences , v. 1, n. 1, p. 83-98, 2008. (Mestrado em Engenharia de Produção) – Instituto Alberto Luiz https://dx.doi.org/10.1504/IJSSCI.2008.017590 Coimbra de Pós-Graduação e Pesquisa de Engenharia, Universidade Federal do Rio de Janeiro, Rio de Janeiro, 1999. SAATY, T.L. How to make a decision: The analytic hierarchy process. European Journal of Operational Research , v. 48, n. 1, p. 9-26, 1990. GOMES, L.F.A.M. TODIM: A System for the Evaluation of Public https://doi.org/10.1016/0377-2217(90)90057-I Transportation Projects. Rio de Janeiro: PUC-RJ, 1987. n. 24. 16 p. SAATY, T.L. The Analytic Hierarchy Process . Nova York: McGraw-Hill, 1980. KAYA, T.; KAHRAMAN, C. An integrated fuzzy AHP–ELECTRE methodology for environmental impact assessment. Expert Systems SEVKLI, M. An application of the fuzzy ELECTRE method for supplier with Applications International Journal, v. 38, n. 7, p. 8553-8562, 2011. selection. International Journal of Production Research , v. 48, n. 12, https://doi.org/10.1016/j.eswa.2011.01.057 p. 3393-3405, 2010. https://doi.org/10.1080/00207540902814355

MARINS, C.S.; SOUZA, D.O.; BARROS, M.S. O uso do método de análise TREVIZANO, W.A.; Freitas, A.P. Emprego do Método da Análise hierárquica (AHP) na tomada de decisões gerenciais: um estudo de Hierárquica (A.H.P.) na seleção de Processadores. In : ENCONTRO caso. In : SIMPÓSIO BRASILEIRO DE PESQUISA OPERACIONAL, 41., NACIONAL DE ENGENHARIA DE PRODUÇÃO, 25., 2005, Porto 2009. Anais [...]. 2009. Alegre. Anais [...]. Porto Alegre, 2005.

| 53 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 43-53 DECISION-MAKING PROCESS

USE OF THE SAPEVO-M METHOD WITH PARAMETERS OF THE SCOR 12.0 MODEL FOR RANKING OF SUPPLIERS IN A SUPPLY CHAIN OF HOSPITAL MATERIAL OF THE BRAZILIAN NAVY Utilização do método SAPEVO-M com parâmetros do modelo SCOR 12.0 para ranqueamento dos fornecedores em uma cadeia de suprimentos de material hospitalar da Marinha do Brasil

Luiz Frederico Horácio de Souza de Barros Teixeira 1, Priscilla Cristina Cabral Ribeiro 2, Carlos Francisco Simões Gomes 3, Marcos dos Santos 4

Abstract: #e study of supply chain management (SCM) can Resumo: O estudo do gerenciamento da cadeia de suprimentos yield bene!ts for all parties involved. A commonly used tool (GCS) pode resultar em benefícios para todas as partes envolvidas. in supply chain (SC) analysis is the Supply Chain Operations Uma ferramenta comumente utilizada nas análises das cadeias de Reference (SCOR), which is a reference diagnostic model for all suprimento (CS) é o Supply Chain Operations Reference (SCOR), steps in a chain. A hospital SC, in turn, has particularities that set um modelo de diagnóstico de referência para todas as etapas de it apart from other areas of industry and commerce. #is paper uma cadeia. Uma CS de material hospitalar, por sua vez, pos- proposes an evaluation of health material suppliers of a Brazilian sui particularidades que a diferenciam de outras áreas da indús- Navy unit through ordinal multi-criteria decision aid method tria e do comércio. Este trabalho propõe uma avaliação de for- SAPEVO-M. #is method is quite recent and little explored necedores de material hospitalar de uma unidade da Marinha do in the literature of this research line. A search in the SCOPUS Brasil por meio do método ordinal multicritério de apoio à deci- database has shown relative novelty in the supply chain corre- são Simple Aggregation of Preferences Expressed by Ordinal Vectors — lation with ordinal methods such as SAPEVO-M used in this Multi Decision Makers (SAPEVO-M). O método é bastante paper. Supplier evaluation was carried out by two experts through recente e pouco explorado na literatura desta linha de pesquisa. the SAPEVO-M online application platform called SapevoWeb. Pesquisa na base Scopus demonstrou relativo ineditismo na cor- #e results achieved allow an analysis of the quality of service relação da cadeia de suprimento com métodos ordinais, como o provided by these companies, as there are no major di%erences SAPEVO-M utilizado neste artigo. A avaliação de fornecedores

1. Master in Production Engineering by Universidade Federal Fluminense. Project Manager of the Naval Systems Analysis Center – Rio de Janeiro, RJ – Brazil. E-mail: [email protected] 2. PhD in Production Engineering by Universidade Federal de São Carlos. Associate Professor in the Department of Production Engineering at Universidade Federal Fluminense (UFF) and Professsor of the Graduate Program in Production Engineering at Universidade Federal Fluminense (TPP/UFF) – Niterói, RJ – Brazil. E-mail: [email protected] 3. PhD in Production Engineering by Universidade Federal do Rio de Janeiro. Associate Professor in the Department of Production Engineering at Universidade Federal Fluminense (UFF) and Professor of the Graduate Program in Production Engineering at Universidade Federal Fluminense (TPP/UFF) – Niterói, RJ – Brazil. E-mail: [email protected] 4. PhD in Production Engineering by Universidade Federal Fluminense. Project Manager of the Naval System Analysis Center and Professor at the Military Engineering Institute – Rio de Janeiro, RJ – Brazil. E-mail: [email protected]

| 54 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos between one alternative and the next. However, this di%erence is foi realizada por dois especialistas, por meio de plataforma evident as they move apart in the ranking. on-line de aplicação do método SAPEVO-M, denominada de Keywords: Supply chain. Supply Chain Operations Reference. SapevoWeb. Os resultados alcançados permitem análise da qua- Multi-criteria decision analysis. Ordinal method SAPEVO-M. lidade do serviço prestado por aquelas empresas, na medida em Brazilian navy. que não há grandes diferenças entre uma alternativa e a seguinte. Contudo, tal diferença se evidencia à medida que as mesmas se distanciam no ranking . Palavras-chave: Cadeia de suprimento. Supply Chain Operations Reference . Métodos multicritérios de apoio à decisão. Método ordinal SAPEVO-M. Marinha do Brasil.

1. INTRODUCTION #e Supply Chain Operations Reference (SCOR) is a model widely used to evaluate the performance of supply When patients are seen in a clinic or hospital, they are chains (SC), although it is generic in nature. For Sellitto et al. unaware of the needs to perform that procedure, however (2015), when approached in the context of decision support, simple it may be. Meanwhile, the process of patient care in the SCOR model is commonly associated with the multi- the hospital is supported by a series of operational activ- criteria Analytic Hierarchy Process (AHP) method. AHP ities, including management and distribution of hospital is used primarily to solve the problem of choosing between supplies to the care sites (MOONS; WAEYENBERGH; alternatives based on a hierarchical structure. PINTELON, 2019). In this article, an evaluation of hospital material sup- Although there are problems faced by public health, the pliers of an MB unit is proposed, using the multi-criteria Armed Forces manage to maintain a quality service to serve ordinal method of decision support Simple Aggregation of their personnel. In the Brazilian Navy ( Marinha do Brazil – Preferences Expressed by Ordinal Vectors – Multi Decision MB), this e%ort occurs through the Navy Health System Makers (SAPEVO-M). It is intended to answer the follow- (NHS). According to the Norms for Medical-Hospital ing research question: how can suppliers of hospital material Assistance of the Navy (BRASIL, 2012), the NHS is the from an MB unit be evaluated? #e focus company will be organized set of human, material, !nancial, technologi- the Navy Procurement Center in Rio de Janeiro ( Centro de cal, and information resources destined to provide health Obtenção da Marinha no Rio de Janeiro – COMRJ), which is activities in the MB. #is system serves the Navy’s mili- the largest material procurement agency for MB, with nation- tary active duty and reserve personnel and their respective wide distribution (BRASIL, 2019c). dependents. Public data from the Statistical Yearbook of the In the present work, the performance metrics of the SCOR Brazilian Navy ( Anuário Estatístico da Marinha do Brasil – model are considered as evaluation criteria of the SC suppli- ANEMAR) (BRAZIL, 2019b) point out that the MB in ers. #e selection of these suppliers and their evaluation were 2018 had more than 210,000 servants, including active and carried out by two specialists, through the SapevoWeb online inactive civilians and military personnel. #us, an e$cient platform (www.sapevoweb.com), speci!c to the application of supply chain management is required in order to provide the SAPEVO-M method. #e assessment consists of paired hospital supplies to this audience. comparisons between the di%erent criteria for the generation Although there are several studies in the literature attest- of weights and between the alternatives in each criterion for ing that supply chain management (SCM) practices can the ordering of suppliers. #e work is divided into !ve sec- result in cost reduction and competitive advantage, health tions. After the introduction, the literature on the SCM and organizations have been too slow to adopt these procedures SCOR model is reviewed and the SAPEVO-M method is (ARONSSON; ABRAHAMSSON; SPENS, 2011). brie]y presented in section 2. Section 3 describes the research

| 55 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos method employed. #en, section 4 details the particular case SCC merged with the American Production and Inventory studied to exemplify the authors’ proposal. #e last section Control Society (APICS), creating APICS SCC, a non-prof- presents the conclusions of this work. itable organization responsible for continuously updating the SCOR model (ASCM, 2017). SCOR combines concepts of process reengineering, bench- 2. LITERATURE REVIEW marking, and application of best practices and allows the organization to detect opportunities to improve its compet- 2.1 SCOR MODEL itiveness. Such a model links business processes, performance #e concept of SC transcends the de!nition of logistics, metrics, practices, and personal skills to a uni!ed structure which, in turn, appears to be inserted in each component of (VELYCHKO, 2015; PEÑA-OROZCO; RIVERA, 2017). the chain in question. According to the Council of Supply #e purpose of adopting standards, such as those provided Chain Management Professionals (CSCMP, 2018), “logis- by the model in question, is to enable companies to develop tics would be the part of the supply chain that basically plans, and maintain processes in their SC that meet certain perfor- implements, and controls the ]ow, storage, and information mance metrics, integrating quality assurance measures into of goods in an e$cient and e%ective manner”. If, previously, the process (LI; SU; CHEN, 2011). #e diagnosis presented logistics was seen as a support activity, often associated only by the model is not restricted only to the evaluation of the with transport and storage, according to Gomes and Ribeiro !nancial performance of the SC, but it is also a support to (2014), it adds value to the functions of customer service, the decision and other aspects of sustainability throughout location, administration, transport, storage, and information the SC (NTABE et al., 2015). Like the other models, SCOR to the productive chain. has metrics for both internal company operations (internal SCM, in turn, would be an evolution of logistics manage- logistics, warehousing, process), and others that go beyond ment, which includes strategic value for allowing companies company boundaries, encompassing from the source to the to achieve better results than they would obtain if they acted delivery processes (MAESTRINI et al., 2017). individually. Such coordination e%ort allows for co-adminis- According to ASCM (2017), the attributes of the SCOR tration of resources, reduction of inventories, and waste and 12.0 model are: maximization of pro!ts (COOPER; LAMBERT; PAGH; • reliability: ability to perform tasks as expected, which 1997). At SCM, companies would seek to reduce operating includes predictability of the outcome of a process; costs, outsourcing some activities and focusing on their core • responsiveness: speed at which tasks are performed and/ ones, which can be di%erential (GOMES; RIBEIRO, 2014). or products are o%ered to customers; In the study of chains, there is concern related to Supply • agility: ability to react to external in]uences and respond Chain Performance Measurement Systems (SCPMS), which to changes in the market to obtain or maintain compet- allows the adoption of performance measurement metrics that itive advantage; a%ect di%erent companies and processes (MAESTRINI et al • costs: price of operating the SC processes, including ., 2017). According to Galo et al. (2018), due to limited com- labor, material, management, and transportation, among munication between SC links, it is possible to think that the others; and measurements are more internal, preventing the adoption of • e$cient asset management: the ability to use assets e$- clear and complete indicators for each criterion and limiting ciently, including inventory reduction, capacity optimi- studies and applications. #us, looking for a more integrated zation, and outsourcing measurement, the SCOR model emerged. It was created by a consulting !rm in 1996 and was soon Research in the Scopus database, carried out on January endorsed by the Supply Chain Council (SCC) as an SCM 24 th , 2020 with the expressions Supply Chain and Supply standard. #is tool allows users to understand the processes Chain Operations Reference obtained, for 2009 to 2020, involved in business organizations and to identify the vital a total of 635 documents, of which 329 articles. #e annual characteristics that lead to customer satisfaction. In 2014, distribution of these articles can be seen in Figure 1.

| 56 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos

Figure 1 has the characteristic sawtooth design, with a #e Chinese predominance can also be seen among the seasonality whose explanation may be associated with the institutions that published the most, with four institutions launch of new models. For example, there was a reduction in among the 13 best placed, as shown in Figure 3. the number of publications in 2009 and an increase from the #e authors who published the largest number of articles end of 2010, which coincides with the launch of the SCOR in the research line are highlighted in Figure 4. It was found 10.0 model in December of that year. #ere is another sharp that there was only one author with prominence in the period: drop in early 2012 and a signi!cant increase in 2013, proba- the Brazilian Luiz Cesar Ribeiro Carpinetti, with six publi- bly because version 11.0 was launched in late 2012. In 2016, cations, which represents twice the number of publications there is a drop, with a further increase in the number of by the eight authors who tied for second place. publications from 2017, the year of launch of SCOR ver- Figure 5 complements this analysis with the pro!le of the sion 12.0. In 2020, there is another sharp drop in the num- journals that had the most articles published in the period. ber of publications, but most likely associated with the fact #ere is also a homogeneous distribution among journals. It is that the data are incomplete because they were compiled by worth mentioning that some of them presented a “sawtooth” January this year. Table 1 illustrates the release dates of the behavior similar to the total annual distribution of articles on most recent SCOR models. the topic, illustrated in Figure 1. Furthermore, there is a coin- As shown in Figure 2, China leads the publications in cidence between the “peak” periods in both graphs. #at is, that period, with 75 articles, followed by the United States, with 36. Brazil appears in the third position in number of Table 1. Release date of the Supply Chain Operations publications, accounting for 24 articles in the period. Reference (SCOR) models. #e leadership among the institutions is divided between the Version Release Chinese Chiang Mai University and the Brazilian University of São Paulo ( Universidade de São Paulo – USP), which con- SCOR 8.0 June / 2006 centrated seven articles each. Another Brazilian institution SCOR 9.0 November / 2008 that stands out is the Federal Technological University of SCOR 10.0 December / 2010 Paraná ( Universidade Tecnológica Federal do Paraná – UTFPR), SCOR 11.0 December / 2012 which shares the second place with the National Kaoshiung University of Science & Technology, in Taiwan, with six SCOR 12.0 October / 2017 publications each. Source: ASCM (2017).

0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Documents/Year

Figure 1. Publication of articles per year in the period 2009–2020.

| 57 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos

China United States Brazil United Kingdom Taiwan Germany France India Malaysia Italy 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Countries/Documents

Figure 2. Distribution of articles by countries.

Chiang Mai University (China) Universidade de São Paulo – USP (Brazil) Universidade Tecnológica Federal do Paraná – UTFPR (Brazil) National Kaoshiung University of Science % Technology (Taiwan) Universiti Utara Malaysia (Malaysia) Huazhong University of Science & Technology (China) Hong Kong Polytechnic University (Hong Kong) National Chiao Tung University (Taiwan) National Yunlin University (Taiwan) National University of Singapure (Singapore) Tianjin University (China) University of Tehran (Iran) Xi’an University of Architecture and Technology (China) 01 2345678 Institution/Documents

Figure 3. Distribution of articles by institutions.

Carpinetti, L.C.R. Chakpitak, N. Huang, L. Lima-Junior, F.R. Panetto, H. Persson, F. Tippayawong, K.Y. Wang, C.N. Xiao, R. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 Authors/Documents

Figure 4. Distribution of articles by authors.

| 58 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos after the launch of a new SCOR model, there is probably a method to multiple decision-makers, in addition to introduc- boom in publications, so journals are more susceptible to that ing a process of normalizing the evaluation matrices, increas- novelty and the total number of articles in the year grows. ing the consistency of the model (TEIXEIRA; SANTOS; During the research on the Scopus database, the authors GOMES, 2019). came across di%erent themes related to the SCOR model. For the same authors, the SAPEVO-M method basically Table 2 presents those most relevant to the survey, while Figure 6 consists of two processes. Firstly, in the ordinal transforma- shows an overview of all subjects. tion of the criteria of preference, expressed by a vector that The Decision Sciences theme is the fourth with the represents the weights of the criteria. #e second process is largest number of associated publications, with 101 articles, the ordinal transformation of preference between alterna- which highlights the relevance between the association of tives in a given set of criteria. #e preference information in Multicriteria Decision Support (MDS) methods and SCM. SAPEVO-M is denoted by a series of paired comparisons between the alternatives. #e relationship between them is 2.2 SAPEVO-M ORDINAL METHOD AND expressed on a seven-point scale, in which the importance of SAPEVOWEB PLATFORM each is relatively measured. Based on the evaluation between In order to solve real problems, Operational Research alternatives, a matrix is obtained with the corresponding adopts logical-mathematical models, characterized by its multidisciplinarity (SANTOS et al., 2015). #e multicrite- Table 2. Themes related to the Supply Chain ria approach to decision support can be characterized as a set Operations Reference (SCOR) model. of methods that shows a problem more clearly, and in which Theme Quantity Percentage the alternatives are evaluated by multiple criteria, which, in Business, Management & 166 21.6% most cases, are con]icting. #ese are techniques for advising Accounting the decision-making agent to make decisions about complex Engineering 153 19.9% problems, with evaluation, ordering or selection of alterna- Computer Science 119 15.5% tives through di%erent points of view and prede!ned criteria (GOMES; GOMES, 2019). Decision Sciences 101 13.2% #e SAPEVO-M method consists of a new version of Social Sciences 45 5.9% the original SAPEVO ordinal method (GOMES; MURY; Environmental Science 39 5.1% GOMES, 1997), which allowed the evaluation of only one Mathematics 28 3.7% decision-making agent. Such an evolved version extends the

6 5 4 3 2 Documents 1 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year

International Journal of Supply Chain Management Production Planning and Control International Journal of Production Economics International Journal of Logistics International Journal of Production Research Research and Applications

Figure 5. Distribution of articles by journals.

| 59 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos

numerical representation. #e preference result is expressed and multi-decision ordinal method of decision support, the by the vector resulting from the matrix multiplication between information transmitted to the decision-making agent is the weight vector of criteria V and the evaluation matrix of not restricted to the provision of alternatives in an order. the alternatives M. #e alternatives are then organized in Additionally, the SapevoWeb software allows to know how decreasing order of the numerical values obtained, generat- much an alternative was better ordered in relation to another ing the desired ranking. (TEIXEIRA; SANTOS; GOMES, 2019). #e computational platform for employing the SAPEVO-M method is called SapevoWeb and can be accessed at: www. sapevoweb.com. #is platform is a program with a friendly 3. RESEARCH METHOD graphic interface that presents results in reduced computational time (TEIXEIRA; SANTOS; GOMES, 2018). #is research started by reviewing the literature on SC #e SapevoWeb system was developed by the Laboratory and the SCOR model, through reading and analyzing articles of Governance, Management and Optimization Studies published in the area. #e authors used the Scopus database, (Laboratório de Estudos de Governança, Gestão e Otimização – accessed through the journals portal of the Coordination for LEGGO), based on a partnership between the technical sta% the Improvement of Higher Education Personnel ( Coordenação of the Naval Systems Analysis Center ( Centro de Análises de de Aperfeiçoamento de Pessoal de Nível Superior – CAPES) on Sistemas Navais – CASNAV), a research group of the Graduate the internet, to consult the most recent and most cited arti- Program in Engineering of Production at the Fluminense cles to prepare the theoretical framework necessary for the Federal University ( Universidade Federal Fluminense – UFF) preparation of this article. and a research group from the Graduate Program in Systems First, on January 24 th , 2020, searches were carried out on and Computer Engineering at the Military Institute of the Scopus website, with the keywords “Supply Chain” and Engineering (Instituto Militar de Engenharia – IME). “Supply Chain Operations Reference”, selecting the option #is platform highlights a novelty introduced by the “All Fields” in the desired search engine !eld. In addition, SAPEVO-M method when expressing numerical values the search was limited, using the database’s “Limit” tool, to arti- associated with ordering when presenting the !nal result. cles published in the period between 2009 and 2020. In this #at is, although SAPEVO-M consists of a multi-criteria !rst stage, 329 articles were obtained, which were presented

Other (7.3%) Agricultural an... (2.0%) Business, Manag... (21.6%) Economics, Econ... (2.6%) Energy (3.3%)

Mathematics (3.7%)

Environmental S... (5.1%)

Social Sciences... (5.9%)

Engineering (19.9%)

Decision Scienc... (13.2%)

Source: adapted from Scopus (2020). Computer Scienc... (15.5%)

Figure 6. Distribution of articles by theme.

| 60 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos according to its number of citations by selecting the option In the same period, searches were made in the Scopus “Sort on: Cited By (highest)”. In the later stage, the articles database with the keywords “SAPEVO” or “SAPEVO-M”, were presented by release date, starting with the most recent, but no document was found related to such methods. through the option “Sort on: Cited By (highest)”. #e authors #erefore, the relative originality of the present work is proven decided, in line with the objective of the work, to conduct a when evaluating suppliers in a SC using the SAPEVO-M statistical survey of the 329 articles found. #is survey is pre- ordinal method. sented in section 2. #e reading of the articles for the theo- In this work, the focus company is COMRJ. #e authors retical foundation was !ltered through the previous reading used public data from the Transparency Portal (BRASIL, 2019) of the titles and, in a second moment, of the abstracts. to select the eight companies with the greatest relevance in the As explained in the previous section, the theme Decision supply of health material for that unit in 2018. To establish Science stands out when associated with SCM, being the the importance of suppliers for this research, it was consid- fourth with the largest number of publications related to SCM ered the total monetary value provided in that year. Figure 7 in the last 10 years. However, the authors decided to verify the represents a section of the SC of hospital material to facili- correlation between the SCOR model and the multi-criteria tate the understanding of its operation and the scope of the decision support methods for typically ordinal problems, whose study. On the left side of the !gure are the main suppliers of intended result is the establishment of a ranking. Research in COMRJ and, on the right side, its main customers. the Scopus database, carried out on January 24 th , 2020 with For a better understanding of the speci!cities of this par- two-by-two combinations between the expressions “supply chain ticular SCM, a military doctor was interviewed (Interviewee operations reference” and “multicriteria ranking”, or “ordinal 1 – I1) and a former businessman who has already worked as multicriteria”, or “ordinal ranking”, obtained only two articles a supplier of hospital supplies for MB (Interviewee 2 – I2). as a result. However, those articles did not apply any method I2 is a retired businessman with market experience in supply- of support for multicriteria decision. Parente et al. (2008) pres- ing this type of material for 35 years, of which he worked for ent an interdisciplinary model for research by SCM, focusing more than ten as a supplier to MB. I1 is an anesthesiologist, on marketing in SC. Ang, Zhu and Yang (2019) integrate the post-graduated in Health Management from the Institute of technique of Stochastic Multicriteria Acceptability Analysis Postgraduate Studies and Research in Administration of the (SMAA) and the methodology of Data Envelopment Analysis Federal University of Rio de Janeiro (COPPEAD/UFRJ), (DEA) to evaluate the e$ciency and classi!cation of SC in head of clinic at Hospital Naval de Brasília and medical o$- two stages with values of stochastic criteria. cer of MB for 20 years. #e interviews were conducted by

đŏ4t!.ŏ+/,%0(.ŏLTDA. %(%0.5ŏ+/,%0( đŏ C+/0ŏ) r#+ŏC+)ċŏ !ŏ .#*%60%+*/ Pr + 1t+/ŏ+/,%0(r!/ŏLTDA. đŏ+/,%0(ŏC!*0r(ŏ đŏ.%/0(%ŏ.+ 10+/ŏ1)% +/ .%*$ŏĨ ĩ F.) n10% +/ŏLTDA. đŏ+/,%0(ŏv(ŏ r (%+ đŏ/,! %f.)ŏC+)ċŏ ! %/ŏĨ Dĩ ! % )!*t+/ŏ!ŏr+ċŏ+/,%0( r!/ŏLTDA. 1,,(%!./ COMRJ (%!*0/ đŏ+/,%0(ŏv(ŏ ! đŏ !0$%+ŏF.) n10% ŏ +ŏr/%(ŏLTDA. r/(%ŏĨĩ đŏr¨ġÒ !ŏ%/0.%1% +r đŏ P+(% (*% ŏv(ŏ+// !ŏ ! % )!*0+/ŏ%.!(%ŏĢŏ !*$+rŏ ŏ(¨.%ŏĨĩŏ đŏr+2% !.ŏ r+ 1t+/ŏ+/,%0(r!/ŏ! đŏ P+(% (*% ŏv(ŏ !ŏY+ *!*t!/ŏLTD ċŏĢŏ P! r+ŏ ŏ( !%ŏĨPĩ đŏ+(1)! ŏ%/0.%1% +.ŏ !ŏ ! % )!*0+/ đŏ2(ŏ+10,0%!*0ŏ (%*% / !ŏr+ 1t+/ŏ,rŏÒ !ŏLt ċŏĢŏ

ME: Individual Microenterprise; EPP: Small Business.

Figure 7. Supply chain (SC) of the Navy Procurement Center in Rio de Janeiro (COMRJ).

| 61 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos telephone and in a free conversation format, in which the risk that institutions run when managing inventory in a SC interviewees were asked about their perception in relation to is the imbalance between demand and supply. In a hospital’s the academic view on SCM found in the literature. SC, in turn, such risk could be mitigated by a$liation with #e question script was made up of open and closed ques- “multi-hospital” system arrangements, mainly local. #e poten- tions. #e open ones raised, as explained above, the under- tial for improvement with such integration can lead to savings standing of the SCM. #e closed ones focused on the criteria of up to 20% of the inventory cost of the integrated health brought from the ASCM publication (2017), that is, on the system in question (ZEPEDA; NYAGA; YOUNG, 2016). attributes of the SCOR 12.0 model: agility (Criterion 1 - C1), #e management of inventory and distribution networks responsiveness (C2), reliability (C3), e$cient management for hospital supplies is among the biggest challenges in the (C4), and cost (C5). #ese criteria were classi!ed in degree of health sector. In fact, stockouts in this sector not only cause importance by a seven-level scale used in the SAPEVO-M !nancial losses, but also put patients’ lives at risk. In addition, method: absolutely worse, much worse, worse, equivalent, many hospitals transfer management practices developed in better, much better, and absolutely better. #e criteria were the industrial sector to the healthcare environment, which evaluated in paired comparisons, as well as the suppliers is not appropriate because of the speci!cities of the sector among themselves, in the light of each one, according to the (KHOUKHI; BOJJI; BENSOUDA, 2019). In this context, scale mentioned above. #e analysis of the data by the rank- MB strives to concentrate and integrate its logistical activities ing was carried out and a re]ection was made of this order for hospital supplies, serving di%erent health units through and how much it represents the assessment and positioning its own supply system, adapted to its context. of suppliers. In addition, a comparison was made between According to the Norms for Supply Execution of the the literature review and the open questions. Brazilian Navy (Brazil, 2009), COMRJ is one of the purchas- In the present work, two military personnel (Decision ing bodies of the Navy Supply System and it is responsible Maker 1 - D1; and Decision Maker 2 - D2) who work in for carrying out research and selection in the market, national the administration of COMRJ were also consulted to carry or foreign, of Providers. In addition, it is also responsible for out the assessment of the main suppliers of health material purchasing items to meet the needs of Military Hospital for that military organization (MO) in 2018. #is assessment Organizations (MHO), primarily those located in the area does not represent the o$cial opinion of MB, serving only for of jurisdiction of the Command of the 1 st Naval District, academic purposes, to exemplify the case study. As criteria, based in the city of Rio de Janeiro. the attributes of the SCOR 12.0 model were used. Decision Hospital SC di%er in several aspects from typical industrial makers 1 and 2 were presented to the de!nitions of these SC, and management practices in sectors such as commerce attributes present in the model. #e evaluation was carried and manufacturing are likely to be a decade ahead of those out directly on the SapevoWeb platform. D1 and D2 were in the healthcare !eld. With the improvement of healthcare encouraged to answer how important they considered each organizations, the increase in operational e$ciency, and the criterion in relation to the others, through comparisons two reduction of costs, the hospital SC is more focused on per- by two; then, similarly, they evaluated the quality of each formance attention (CHEN; PRESTON; XIA, 2013). alternative in each criterion, in comparison to the others. In 2018, according to the Transparency Portal, the amount of health material purchased by COMRJ to be transferred to MHO or medical facilities it serves was more than R$ 10 4. CASE STUDY: HOSPITAL million (BRASIL, 2019a). In addition, hospitals manage large MATERIAL SUPPLY TO THE quantities and varieties of items and the storage and distribu- BRAZILIAN NAVY tion of all this material are essential for providing high quality services to the patient (MOONS; WAEYENBERGH; In inventory management, it is relevant to observe actions PINTELON, 2019). related to policies, administration, and inventory control Aronsson, Abrahamsson and Spens (2011) question why the (GOLOBOVANTE; RIBEIRO; GOMES, 2018). #e main delay in the managing practices of the SC of health material,

| 62 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos considering that 30 to 40% of hospital expenses are invested Figures 8 and 9 illustrate the use of the SapevoWeb online in activities related to logistics. According to E2, an obstacle system for, respectively, the evaluation of criteria and alter- for public hospitals in Brazil would be excessive bureaucracy. natives. #e !rst step in using the SapevoWeb software is to As in any other public administration body, all material must insert the problem data by the user. After inserting the criteria be purchased through tenders, which are generally too slow. to be evaluated, the system generates a comparative evalua- Volland et al. (2017) attest that hospital SC is often tion page between them, two by two, for each previously regis- impaired, as health professionals are diverted from care tered decision-making agent. At the end of the evaluation, the for logistical activities for which they were not prepared. SapevoWeb system generates a comparative evaluation page I1 refutes this position, understanding that this would not between the alternatives in each criterion, similarly to the pre- be a problem for the Navy, which is concerned with train- vious process. After comparing the alternatives, the evaluations ing its doctors in business management, as they move up of each decision-making agent were inserted, and the system to higher positions in their careers. #e same interviewee instantly generated a new page with the results obtained from emphasizes that excellence in service provision is sought the weight value of the criteria and the ordering of the alter- with a focus on governance, increasingly aiming at e$- natives, based on the SAPEVO-M method. ciency and e%ectiveness in processes. #erefore, for him, #e SapevoWeb system itself calculates and presents the it is important to have managers in the health !eld who results: weight of the criteria, which represents their relative know the activity well, and it is increasingly common for importance to the evaluators; and the degree obtained by medical assistants to also go to administrative functions. the alternatives, in decreasing order of score. Table 3 indi- #us, for I1, such professionals are able to have a global cates the value of the criteria weights calculated by the sys- view of the subject and make better decisions for the insti- tem, while the !nal result of the ranking is shown in Table 4. tution, professionals, and patients. Table 3 shows that the consulted COMRJ administrators In the present case study, decision makers D1 and D2, favor, mainly, the reliability criterion of the SC of hospital who work in the administration of COMRJ, were questioned material. Probably, as they are health materials, it is import- about their perception of the relative importance of the cri- ant for those decision makers to be sure that the agreements teria, taken from the attributes of the SCOR 12.0 model. are complied with so that care is not compromised. #en, the same decision-makers were asked to evaluate the Table 4 shows that there were no major discrepancies performance of the eight suppliers selected under each criterion. between the degrees obtained by a given company and that

Source: SapevoWeb (2020). Figure 8. Extract from the criteria evaluation page.

| 63 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos

subsequently ranked. In the case studied, the close notes may Table 3. Criteria weight. mean that there are no signi!cant di%erences between the Attribute Weight Criterion service provided by a company and that which follows in Agility 1.5000 C1 the 2018 supplier ranking. On the other hand, as the result is presented on an absolute scale, one can observe that the !rst Responsiveness 0.7923 C2 place had a performance close to double that of the seventh Reliability 1.9231 C3

place, or that the eighth place had a performance equivalent Eocient Asset Management 0.0056 C4 to half of the second place that year. #is means that, if there Cost 0.8615 C5 are no major di%erences between one alternative and the next, that di%erence becomes evident as they move away in the ranking. #is information can be essential for a decision Table 4. Ranking of alternatives: suppliers. maker, for example, to understand the gains and losses in the Ranking Supplier Degree obtained selection of a certain alternative over another. In a hypothetical situation in which a supplier better placed in the ranking 1st CRISTALIA LTDA. 6.58253 cannot be selected for any reason outside the will of the deci- 2nd BAXTER LTDA. 5.46576 sion-making agent, he could have greater security to analyze 3rd COSTA LTDA. 5.36367 the selection of other alternatives in the ranking. 4th ESPECIFARMA 4.96107

5th SOLUMED 4.64744

5. CONCLUSION 6th PRO-SAUDE 3.91016

7th PROVIDER 3.38812 #e SapevoWeb system emerges as an important tool 8th METHABIO 2.65502 for easily obtaining results from complex problems and for

Source: SapevoWeb (2020). Figure 9. Extract from the alternative evaluation page for each criterion.

| 64 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos researchers to focus on other stages of the process. #e free certain item, customers could accept the second place in the online availability of this multicriteria decision aid platform ranking as a viable alternative, without signi!cant damage makes it possible to disseminate a new ordinal method to to the quality of service. On the other hand, if the option the academic community. were placed at the bottom of the ranking, customers could The present study proposed to use the MDS ordi- see a drop in the quality of the service presented, consid- nal method from a di%erent perspective in a SCM study, ering that this alternative scored 2.5 times lower than the introducing a new possibility of approach to value SCM company that ranked !rst in the evaluation. aspects through ordering. According to the experts con- To improve the management of hospital supplies at MB, sulted, the company Cristalia LTDA. was the best placed it is suggested that this research be replicated with an even in the ranking of eight hospital suppliers from COMRJ. greater number of suppliers, the employment of more spe- However, when presenting a grade for each ordered alter- cialists from COMRJ as evaluators and extension to other native, the SAPEVO-M method allows for subjective sectors of the NHS. Future research can carry out a literature analysis of the di%erence in service quality of each com- review in other databases, use the metrics of the SCOR model pany. Consequently, in a hypothetical situation in which as criteria to evaluate suppliers or extend the application of that company could not compete to be selected to supply a the SAPEVO-M method to other levels of the supply chain.

REFERENCES

ANG, S.; ZHU, Y.; YANG, F. Eciency evaluation and ranking of supply chain performance: A relational view and empirical test. Journal of chains based on stochastic multicriteria acceptability analysis and Operations Management , v. 31, n. 6, p. 391-408, 2013. https://doi. data envelopment analysis. International Transactions in Operational org/10.1016/j.jom.2013.07.012 Research , p. 1-30, 2019. http://doi.org/10.1111/itor.12707 COOPER, M. C.; LAMBERT, D. M.; PAGH, J. D. Supply Chain ASSOCIATION FOR SUPPLY CHAIN MANAGEMENT (ASCM). APICS Management: More Than a New Name for Logistics. The International SCC . About APICS SCC. 2017. Disponível em: . Acesso em: 29 nov. 2018. org/10.1108/09574099710805556

ARONSSON, H.; ABRAHAMSSON, M.; SPENS, K. Developing lean and COUNCIL OF SUPPLY CHAIN MANAGEMENT PROFESSIONALS agile health care supply chains. Supply Chain Management, v. 16, n. 3, (CSCMP). Disponível em: . Acesso em: 24 nov. 2018. p. 176-183, 2011. https://dx.doi.org/10.1108/13598541111127164 GALO, N. R.; RIBEIRO, P. C. C.; MERGULHÃO, R. C.; VIEIRA, J. G. V. BRASIL. Controladoria Geral da União. Portal da Transparência . Disponível Selección de proveedor de servicios logísticos: alineación entre em: . Acesso em: 12 jul. 2019a. criterios e indicadores. Innovar , v. 28, n. 69, p. 55-70, 2018. http://

BRASIL. Ministério da Defesa. Marinha do Brasil. Anuário Estatístico dx.doi.org/10.15446/innovar.v28n69.71696 da Marinha do Brasil (ANEMAR) 2018 . Rio de Janeiro: Secretaria-Geral GOLOBOVANTE, A. F. M.; RIBEIRO, P. C. C.; GOMES, C. F. S. Adoção de da Marinha, Diretoria de Administração da Marinha, 2019b. v. 1. Resposta Rápida e gerenciamento de estoques no fast fashion: dois

BRASIL. Ministério da Defesa. Marinha do Brasil. Carta de Serviços estudos de caso em Minas Gerais. Gestão & Produção , v. 25, n. 3, p. ao Cidadão . Rio de Janeiro: Centro de Obtenção da Marinha no 557-567, 2018. https://doi.org/10.1590/0104-530x2063-18

Rio de Janeiro, 2019c. Disponível em: . integrada à Tecnologia da Informação . 2. ed. Rio de Janeiro: Editora Acesso em: 4 set. 2019. Senac, 2014. BRASIL. Ministério da Defesa. Marinha do Brasil. Normas para GOMES, L. F. A. M.; GOMES, C. F. S. Princípios e métodos para a tomada Assistência Médico-Hospitalar da Marinha do Brasil (DGPM-401) . de decisão: enfoque multicritério. 6. ed. São Paulo: Atlas, 2019. Rio de Janeiro: Diretoria Geral de Pessoal Militar, 2012. GOMES, L. F. A. M.; MURY, A. R.; GOMES, C. F. S. Multicriteria ranking BRASIL. Ministério da Defesa. Marinha do Brasil. Normas para with ordinal data. Systems Analysis Modelling Simulation , 1997. Execução do Abastecimento da Marinha do Brasil (SGM - 201) . Rio de Janeiro: Secretaria Geral da Marinha, 2009. KHOUKHI, S.; BOJJI, C.; BENSOUDA, Y. A review of medical distribution logistics in pharmaceutical supply chain. International Journal of CHEN, D. Q.; PRESTON, D. S.; XIA, W. Enhancing hospital supply Logistics Systems and Management , v. 34, n. 3, p. 297-326, 2019.

| 65 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66 Luiz Frederico Horácio de Souza de Barros Teixeira, Priscilla Cristina Cabral Ribeiro, Carlos Francisco Simões Gomes, Marcos dos Santos

LI, L.; SU, Q.; CHEN, X. Ensuring supply chain quality performance through SELLITTO, M. A.; PEREIRA, G. M.; BORCHARDT, M.; SILVA, R. I. da; applying the SCOR model. International Journal of Production Research , VIEGAS, C. V. A SCOR-based model for supply chain performance v. 49, n. 1, p. 33-57, 2011. http://doi.org/10.1080/00207543.2010.508934 measurement: application in the footwear industry. International Journal of Production Research , v. 53, n. 16, p. 4917-4926, 2015. https:// MAESTRINI, V.; LUZZINI, D.; MACCARRONE, P.; CANIATO, F. Supply dx.doi.org/10.1080/00207543.2015.1005251 chain performance measurement systems: A systematic review and research agenda. International Journal of Production Economics , TEIXEIRA, L. F. H. S. B.; SANTOS, M.; GOMES, C. F. S. SapevoWeb v. 183, parte A, p. 299-315, 2017. https://doi.org/10.1016/j.ijpe.2016.11.005 Software: v. 1. 2018. Disponível em: . Acesso em: 30 jul. 2019. MOONS, K.; WAEYENBERGH, G.; PINTELON, L. Measuring the logistics performance of internal hospital supply chains –A literature TEIXEIRA, L. F. H. S. B.; SANTOS, M.; GOMES, C. F. S. Proposta e study. Omega, v. 82, p. 205-217, 2019. implementação em python do método Simple Aggregation of Preferences Expressed by Ordinal Vectors - Multi Decision Makers : NTABE, E.N.; LEBEL, L.; MUNSON, A.D.; SANTA-EULALIA, L.A. A uma ferramenta web simples e intuitiva para Apoio à Decisão systematic literature review of the supply chain operations reference Multicritério. In : SIMPÓSIO DE PESQUISA OPERACIONAL E (SCOR) model application with special attention to environmental LOGÍSTICA DA MARINHA, 19., 2019, Rio de Janeiro, RJ. Anais […]. Rio issues. International Journal Production Economics , v. 169, p. 310-332, de Janeiro: Centro de Análises de Sistemas Navais, 2019. 2015. https://doi.org/10.1016/j.ijpe.2015.08.008 VELYCHKO, O. Integration of SCOR-Modeling and Logistical Concept PARENTE, D. H.; LEE, P. D.; ISHMAN, M. D.; ROTH, A. V. Marketing and of Management in the System of Internal Transportation of Milk supply chain management: a collaborative research agenda. Journal Cooperative. Mediterranean Journal of Social Sciences , v. 6, n. 1, of Business & Industrial Marketing , v. 23, n. 8, p. 520-528, 2008. https:// supl. 2, p. 14-24, 2015. http://doi.org/10.5901/mjss.2015.v6n1s2p14 doi.org/10.1108/08858620810913335 VOLLAND, J.; FÜGENER, A.; SCHOENFELDER, J.; BRUNNER, J. O. PEÑA-OROZCO, D. L.; RIVERA, L. Sensitivity analysis of the SCOR Material logistics in hospitals: A literature review. Omega, v. 69, p. metrics selected for the measurement of the management of a fruit- 82-101, 2017. http://dx.doi.org/10.1016/j.omega.2016.08.004 growing supply chain. Dyna, v. 84, n. 203, p. 306-315, 2017. https:// dx.doi.org/10.15446/dyna.v84n203.61442 ZEPEDA, E. D.; NYAGA, G. N.; YOUNG, G. J. Supply chain risk management and hospital inventory: Eects of system aliation. SANTOS, M.; QUINTAL, R. S.; PAIXÃO, A. C. da; GOMES, C. F. S. Simulation Journal of Operations Management , v. 44, p. 30-47, 2016. of Operation of an Integrated Information for Emergency Pre-hospital Care in Rio de Janeiro Municipality. Procedia Computer Science , v. 55, p. 931-938, 2015. https://dx.doi.org/10.1016/j.procs.2015.07.111

| 66 | Revista Pesquisa Naval, Rio de Janeiro - RJ, n. 31, 2019, p. 54-66