Emergency evacuation in high-rise apartment building is a challenge for fire safety professionals. Lift evacuation is a controversial issue because the safe operation of lift is not ensured under the existing design and operating conditions. Lifts are not permitted for public evacuation in apartment buildings during fire emergencies as per the provisions of building codes and regulations. However, the concept of using lifts for emergency evacuation has been gaining considerable attention during recent years. The lift evacuation can be considered as an alternative facility if it is efficient, reliable and readily accessible. It can also provide a safer means of evacuation for the aged and disabled persons, who may not be able to evacuate promptly, efficiently and unassisted using the exit stairs during fire emergencies. Moreover, lifts can enable building corporate management to easily and promptly access the fireaffected floor and commence fire fighting. The work on the use of lifts for emergency evacuation was initiated in the early 1990s at the National Institute of Standards and Technology (NIST, USA) in which pros and cons were analysed in order to develop suitable guidelines. This research project examines the feasibility of using lifts along with design modifications as an alternative facility for a safer and more efficient emergency evacuation. The scope of this research is limited to apartment buildings where occupant load is low and fire load is generally confined to dwelling compartment units. This research project analysed the important issues in relation to the use of lifts for emergency evacuation. The issues were divided into three categories: human behavioural response, fire hazards and lift operational mechanism. Output variables relating to human behavioural response were modelled and analysed as a stochastic process. Residents’ choice for using evacuation routes was determined using a survey. The issues of fire hazards (fire, smoke and toxic gases) were analysed for occupant safety under variable conditions using the concept of fire safety index. The issues of lift operational mechanism such as lift malfunctioning due to excessive temperature, electric power failure and water damage were considered for developing probabilistic models. An integrated approach of risk assessment for the issues of human behavioural response and fire hazards (such as ‘decision uncertainty’, ‘panic’, ‘nonfatal and fatal injuries’) was developed based on the Multi-Objectives Decision Analysis method. The results for lift and stair systems were compared and the feasibility of using lift with design modifications was analysed for alternative designs and evacuation strategies. The outcomes of this research have shown that using lifts with a protected lobby for up to one-fourth of the building population (who may be aged and disabled) has huge potential as an alternative evacuation facility with enhanced level of safety. Lifts with protected lobby for one-fourth of the building population showed an improved level of fire safety from exposure to fire effluents. The reliability of lift operational mechanism is also improved in protected lift shafts. Lifts with protected lobby for up to one-fourth of the building population and stairs for up to three-fourth of the building population showed an improved evacuation safety. The risks in combined evacuation systems (protected lifts and stairs) are found to be lower when compared to using stairs or protected lifts. Lifts with double lobby protection (for example, two levels of compartmentation with fire and smoke doors for lift lobby) showed further improvements. This research has proposed alternative designs for lifts and developed models for analyzing evacuation effectiveness based on risks related to human behaviour, fire hazards and operational mechanism. It has shown that a combined use of lifts and stairs has significant advantages. The performance based lift evacuation system is achievable in apartment buildings. These research findings are based on uncertainty analysis, which can be further extended to other types of buildings in the future.

No description available.

Website hosts want interested visitors who engage in the activities that are the purpose of a website. This is usually achieved by designing the website to be simple to navigate and aesthetically pleasing. As the design is done before the visits, it would be an exciting addition if the website could identify the visitors who become disinterested during their visits, and then offer personalized motivation for the visitor to engage in the website. This study aims to identify whether a visitor is about to leave a website, by using machine learning models to predict their exits. The purpose of that is to offer personalized motivation in real time for visitors to continue their visits. That is however outside the scope of this study. This research investigates how well machine learning models can predict website exits from session data. The algorithms chosen for the prediction are an Artificial Neural Network (ANN) and a Support Vector Machine (SVM). These are trained on session data. An important part of the research is to extract suitable features from the session data to enhance the prediction. The models are cross-validated and it is found that the models show success in predicting exits, and that it is possible to predict exits with a performance of at least 0.70 Area Under the Curve (AUC). / Webbplatsvärdar vill gärna ha intresserade besökare som engagerar sig i aktiviteterna som är syftet med en webbplats. Det åstadkoms vanligtvis genom att designa webbplatser som är lätta att navigera på och estetiskt tilltalande. Då design görs innan besöken vore det ett spännande tillägg om webbplatsen kunde identifiera besökare som blir ointresserade under deras besök, och erbjuda dem personlig motivation för att engagera sig i webbplatsen. Den här studien försöker identifiera huruvida en besökare är på väg att lämna en webbplats, genom att använda maskininlärningsmodeller för att förutse deras avbrott i besöken. Syftet med det är att i realtid erbjuda dem personlig motivation att fortsätta deras besök. Men just det är utanför den här studiens omfång. Den här studien undersöker hur väl maskininlärningsmodeller, baserat på sessionsdata, kan förutse avbrotten i webbplatsbesök. De valda algoritmerna för prediktion är ett Artificiellt Neuralt Nätverk (ANN) och en Stödvektorsmaskin (SVM). Dessa tränas på sessionsdata. Det en viktig del av studien att utvinna egenskaper ur sessionsdata för att förbättra prediktionen. Modellerna kors-valideras och det upptäcks att modellerna visar framgång i att förutse avbrott, och att det är möjligt att förutse avbrott med en prestanda på åtminstone 0.70 Area Under Kurvan (AUC).

A segurança contra incêndios ainda é um tema incipiente no Brasil. Em São Paulo a primeira legislação data de 1983, sendo conseqüência das tragédias ocorridas nos anos de 1970, como a do Edifício Andraus, em 1972, e a do Edifício Joelma em 1975. Este trabalho procura discorrer sobre o projeto de saídas de emergência dos edifícios escolares construídos pelo Estado de São Paulo e em especial, os mais contemporâneos, que estão elevando seu gabarito de altura e concentrando sua população. Inicialmente fez-se uma abordagem histórica onde procurou-se examinar as principais características tipológicas das escolas construídas pelo Estado de São Paulo e, no capítulo seguinte, um levantamento das legislações edilícias produzidas pelo Município e pelo Estado de São Paulo, sempre com foco na segurança contra incêndio e, em especial, nas saídas de emergência. Em seguida, procurou-se compreender o conceito de desempenho, os principais modelos matemáticos utilizados na segurança contra incêndios e as variáveis que podem influenciar um processo de abandono, tais como: a velocidade das pessoas, o tempo de resposta ao alarme e o comportamento humano em situações de incêndio. Por fim, algumas simulações de abandono por computador para quatro escolas foram realizadas, utilizando o modelo matemático denominado buildingExodus, com o objetivo de avaliar as saídas de emergência e contribuir para o aprofundamento e a divulgação do conhecimento sobre o assunto. / Fire safety is still an incipient subject in Brazil. In São Paulo, the first piece of legislation dates back to 1983, being a consequence of the tragedies that occurred in the 1970s, such as the one in the Andraus Building in 1972 and the one in the Joelma Building in 1975. This work aims at discoursing on the design of the emergency exits of the school buildings that have been constructed by the State of São Paulo, especially the design of the most contemporary structures, which have been increasing in height and concentrating their population. Initially, a historical perspective is presented, followed by an examination of the principal characteristics of the schools that the State of São Paulo has built and by a study of the building regulations that the City and the State of São Paulo have created, always focusing on fire safety and, especially, on emergency exits. After that, this work tries to understand the performance-based concept, the principal mathematical models utilized in fire safety and the factors that may influence an evacuation process, such as: peoples travel speeds, the time delay to start evacuating on hearing a fire alarm, and the human behavior in fire situations. Finally, some computer-based evacuation simulations in four schools were done, utilizing the mathematical model called buildingExodus, in order to evaluate the emergency exits and to contribute to the further understanding and dissemination of the subject.

A segurança contra incêndios ainda é um tema incipiente no Brasil. Em São Paulo a primeira legislação data de 1983, sendo conseqüência das tragédias ocorridas nos anos de 1970, como a do Edifício Andraus, em 1972, e a do Edifício Joelma em 1975. Este trabalho procura discorrer sobre o projeto de saídas de emergência dos edifícios escolares construídos pelo Estado de São Paulo e em especial, os mais contemporâneos, que estão elevando seu gabarito de altura e concentrando sua população. Inicialmente fez-se uma abordagem histórica onde procurou-se examinar as principais características tipológicas das escolas construídas pelo Estado de São Paulo e, no capítulo seguinte, um levantamento das legislações edilícias produzidas pelo Município e pelo Estado de São Paulo, sempre com foco na segurança contra incêndio e, em especial, nas saídas de emergência. Em seguida, procurou-se compreender o conceito de desempenho, os principais modelos matemáticos utilizados na segurança contra incêndios e as variáveis que podem influenciar um processo de abandono, tais como: a velocidade das pessoas, o tempo de resposta ao alarme e o comportamento humano em situações de incêndio. Por fim, algumas simulações de abandono por computador para quatro escolas foram realizadas, utilizando o modelo matemático denominado buildingExodus, com o objetivo de avaliar as saídas de emergência e contribuir para o aprofundamento e a divulgação do conhecimento sobre o assunto. / Fire safety is still an incipient subject in Brazil. In São Paulo, the first piece of legislation dates back to 1983, being a consequence of the tragedies that occurred in the 1970s, such as the one in the Andraus Building in 1972 and the one in the Joelma Building in 1975. This work aims at discoursing on the design of the emergency exits of the school buildings that have been constructed by the State of São Paulo, especially the design of the most contemporary structures, which have been increasing in height and concentrating their population. Initially, a historical perspective is presented, followed by an examination of the principal characteristics of the schools that the State of São Paulo has built and by a study of the building regulations that the City and the State of São Paulo have created, always focusing on fire safety and, especially, on emergency exits. After that, this work tries to understand the performance-based concept, the principal mathematical models utilized in fire safety and the factors that may influence an evacuation process, such as: peoples travel speeds, the time delay to start evacuating on hearing a fire alarm, and the human behavior in fire situations. Finally, some computer-based evacuation simulations in four schools were done, utilizing the mathematical model called buildingExodus, in order to evaluate the emergency exits and to contribute to the further understanding and dissemination of the subject.

Mehr als zwei Drittel der Menschen unterhalb der Armutsgrenze in Sub-Sahara Afrika leben im ländlichen Raum; vier Fünftel von ihnen erwirtschaften ihren Lebensunterhalt überwiegend durch landwirtschaftliche Aktivitäten. Somit spielt die Landwirtschaft eine herausragende Rolle in der ländlichen Wirtschaftsentwicklung und Armutsbekämpfung. Dem Sektor kam in der letzten Dekade wieder gestiegene Aufmerksamkeit durch Ent-wicklungsforschung und Entwicklungszusammenarbeit zu. Jedoch ist der Zusammen-hang zwischen landwirtschaftlichen Aktivitäten und ihrem Beitrag zur Überwindung der Armut auf der Haushaltsebene („poverty exit”) kaum empirisch analysiert. In Kenia trägt der Agrarsektor 26 % zum BIP bei, vor- und nachgelagerten Bereiche zusätzliche 27 %. Mehr als 70 % der ländlichen Bevölkerung arbeiten in der Landwirt-schaft; landesweit sind das mehr als 40 % der Gesamtbevölkerung. Seit der Jahrtau-sendwende ist der Sektor im Durchschnitt um 3 % gewachsen, allerdings mit hoher Varianz. Seitdem haben sich viele kleinbäuerliche Haushalte in landwirtschaftliche Wert-schöpfungsketten wie Exportgemüse oder Milch integriert. Gleichzeitig sank die ländli-che Armutsrate von 49,7 % auf 40,1 %. Was also war der Beitrag der Wertschöpfungs-kettenentwicklung auf die ländliche Armutsreduktion? Diese Arbeit untersucht am Beispiel Kenias, warum manche ländlichen Haushalte die Armut überwunden haben und welche landwirtschaftlichen Aktivitäten dazu geführt haben. Die Basis dafür bilden drei unterschiedliche Stränge in der Literatur: die Pro-Poor Growth-Debatte, die Wertschöpfungskettenliteratur, und die Armutsforschung entlang des sogenannten „Q-Squared Paradigms“. Elemente dieser drei Stränge bilden das analy-tische Gerüst für die empirische Analyse im ländlichen Kenia und kombinieren quanti-tative und qualitative Methoden. Zuerst wurde das ländliche Haushaltpanel „TAMPA“ mit 1275 ländlichen Haushalten über einen Zeitraum von 10 Jahren (1997-2007) nach Haushalten ausgewertet, die die Armut überwunden haben. Eine Stichprobe von 51 dieser Haushalte wurde im Jahr 2010 nachbefragt, um die spezifischen Gründe für den wirtschaftlichen Aufstieg zu analysieren. Die Ergebnisse dieser Befragung lassen sich auf zwei Ebenen interpretieren und leisten daher zwei unterschiedliche Beiträge zur ländlichen Armutsforschung: erstens methodi-sche Hinweise zur Datenerfassung und -interpretation ländlicher Haushaltspanele in Afrika; zweitens Erkenntnisse zu effektiven Strategien, wie die Integration in landwirt-schaftliche Wertschöpfungsketten zur Überwindung ländlicher Armut führen kann. Das zentrale methodische Ergebnis liegt darin, dass obwohl die quantitativen Daten von vier Haushaltsbefragungen im Rahmen des Panels eindeutig eine Überwindung der Armut bei allen 51 Haushalten zeigen, nur 25 dieser Haushalt tatsächlich der Armut entkommen sind. Die anderen 26 Haushalte haben sich unterschiedlich entwickelt oder sind Messfehlern unterlegen. Die Lebensgeschichten („life histories“) der 25 Haushalte, die die Armut überwunden hatten, zeigen jedoch eindeutig, dass die Integration in landwirtschaftliche Wertschöpfungsketten tatsächlich einen sehr guten Entwicklungs-pfad aus der Armut bieten kann, wenn die landwirtschaftlichen Aktivitäten produktiv und marktorientiert sind und Investitionen und Innovationsadoption erfahren. Die Mischung aus Spezialisierung auf eine kommerzielle Wertschöpfungskette, die Elemente kollektiver Vermarktungsstrukturen wie Kooperativen aufweist, einerseits und risikomi-nimierender Diversifizierung inklusive ernährungsrelevantem Eigenkonsum anderer-seits, scheint hierbei der erfolgreichste Weg aus der ländlichen Armut zu sein. Beide Ergebnisbereiche führen zu Schlussfolgerungen, wie zukünftig landwirtschaftliche Wertschöpfungskettenförderung den Fokus auf Armutsreduzierung stärken kann und wie die Armutsforschung die Umsetzung solcher Projekte und deren Armutseffekte besser messen könnte.:1 INTRODUCTION 2 POVERTY AND GROWTH DEBATES AND THEIR IMPLICATIONS FOR AID TO AGRICULTURE 3 POVERTY DYNAMICS AND POVERTY EXITS: CONCEPTS, METHODOLOGIES AND CASES STUDIES 4 EMPIRICAL EVIDENCE: POVERTY EXITS IN RURAL KENYA 5 CONCLUSIONS AND RECOMMENDATIONS / In Sub-Saharan Africa, more than two thirds of the poor live in rural areas and four out of every five households are making a living predominantly based on agricultural and livestock activities. Agriculture plays a major role in stimulating rural economic development and in driving rural households out of poverty. Hence, the sector experienced a revival in development cooperation and devel-opment research during the past decade. However, it is rarely analysed which agricultural strategies rural households undertake to successfully exit poverty. In Kenya, agriculture is key to the economy, contributing 26 % of the GDP di-rectly and another 27 % indirectly. The sector provides employment for more than 40 % of the total population and more than 70 % of the rural population. Since the Millennium, the sector experienced an average growth rate of 3 % per year, albeit high variances. Since then, many small-scale farmers in Kenya have integrated their economic activities into agricultural value chains, such as food crops, export horticulture or dairy. At the same time, rural poverty incidence re-duced from 49.9 % to 40.1 %, but is still significantly higher than urban poverty. So what was the impact of value chain development on rural poverty reduction? This research analyses why some rural households exited poverty and to what extent these poverty exits are explained by their agricultural activities. Based on a literature review of the pro-poor growth debate, of agricultural value chain de-velopment and of poverty research along the ‘q-squared-paradigm', the results from three different schools are combined for the design of an empirical field survey in rural Kenya applying quantitative and qualitative methods. First, the ten-year TAMPA panel data set for 1275 rural households was analysed for the identification of poverty exiting households. These poverty exiters were then sampled again for qualitative follow-up interviews in order to specifically analyse their explanation for their upward mobility. 51 households were visited and in-terviewed for their agricultural life history in 2010. The results are two-fold: first, even though the four wave panel data for all 51 households showed a clear upward trend, only 25 households turned out to have actually exited poverty between 1997 and 2007. The other 26 households had either never been poor or were still poor, or had progressed in their lifecycle and remaining resources were divided by fewer dependants. Thus, a number of con-clusions are drawn for the interpretation and further use of such panel data. Sec-ond, the interviews with ‘real’ poverty exiters confirm that the integration into agricultural value chains can offer a stable pathway out of poverty, if the agricul-ture and livestock portfolio of the households is productive, receives invest-ments and innovation, is commercially oriented and linked to markets. Against the common notion that specialisation in few activities usually marks this neces-sary productivity, here, a combined specialised and diversified pathway is ob-served to be most successful. Agricultural value chain development with a focus on horizontal cooperation and collective marketing of cash crops or dairy in combination with a diversified food crop portfolio seems to have been the most promising pathway out of rural poverty. Both result areas provide recommendations for the implementation of future ag-ricultural value chain projects as well as for future rural poverty research.:1 INTRODUCTION 2 POVERTY AND GROWTH DEBATES AND THEIR IMPLICATIONS FOR AID TO AGRICULTURE 3 POVERTY DYNAMICS AND POVERTY EXITS: CONCEPTS, METHODOLOGIES AND CASES STUDIES 4 EMPIRICAL EVIDENCE: POVERTY EXITS IN RURAL KENYA 5 CONCLUSIONS AND RECOMMENDATIONS

No description available.

Made available in DSpace on 2016-04-28T20:38:06Z (GMT). No. of bitstreams: 1 Cinthia de Carvalho Lourenco.pdf: 1539301 bytes, checksum: 29ed2eb1e4e5cf5ecfa5fb371c34370a (MD5) Previous issue date: 2012-05-07 / Conselho Nacional de Desenvolvimento Científico e Tecnológico / In the context of Post-Modernity society has been facing numerous changes in the families shapes. One of them is the attempt of a young adult to leave and, later, return to the parental home. In order to understand what motivates such phenomenon, a qualitative research including case study was made. The theoretical assumptions underlining this paper are researches and literature discussing the subject, as well as authors of the Family Systems Theory, but aiming for a differentiated interpretation. The analysis of the interviews was conducted under two guidelines: differentiation and general aspects of the exits and returns of the son to the parents house, a summary about the family s life history and a genogram. The research provided data showing the cycle of leaving and coming back to parent s house is related to certain aspects of a family dynamics, to the relationship between parents and sons and social and economic characteristics of our contemporary reality. It was noted there are few researches about the subject, especially related to its psychological aspects. Data found in current literature refers mainly about the prolonged stay of young adult with his/her parents. Therefore, it is necessary to further research on the topic in question / No contexto da Pós-Modernidade deparamo-nos com inúmeras modificações nos contornos familiares. Uma delas é a tentativa dos filhos de saírem e retornarem ao lar parental. Buscando compreender quais fatores motivam esse fenômeno foi realizada uma pesquisa qualitativa com delineamento de estudo de caso. Os pressupostos teóricos utilizados neste trabalho envolveram pesquisas e literatura que abordam o tema, bem como os autores da Teoria Sistêmica, com vistas no processo de diferenciação. A análise do conteúdo das entrevistas foi realizada em torno de dois eixos: diferenciação e aspectos gerais das saídas e retornos do filho à casa dos pais, elaborado um resumo sobre a história de vida dessa família e construído seu genograma. O conhecimento gerado demonstrou que as idas e vindas à casa dos pais relacionam-se aos aspectos da dinâmica familiar, ao vinculo e relacionamento estabelecido entre pais e filha, e as características do contexto socioeconômico da contemporaneidade. Averiguou-se que há pouco registro de estudos sobre o assunto, principalmente no que se refere aos aspectos psicológicos que o permeiam. Os dados encontrados na literatura atêm-se ao prolongamento dos filhos na casa dos pais, portanto, fazem-se necessárias novas pesquisas acerca do tema em questão

Made available in DSpace on 2010-04-20T20:51:24Z (GMT). No. of bitstreams: 3 61060100565.pdf.jpg: 20837 bytes, checksum: ecb09c5a7fdd6fe526137a430cca9641 (MD5) 61060100565.pdf: 372772 bytes, checksum: fe742ffcf601cb252e6a7c1dad0561bf (MD5) 61060100565.pdf.txt: 165929 bytes, checksum: 2a480140de7c74b3cdb7fab1a1d36eca (MD5) Previous issue date: 2008-02-13T00:00:00Z / This paper investigates the influence of the characteristics of private equity and venture capital (PE/VC) organizations on the performance of their funds. Specifically, the analysis seeks to detect how the profiles of investment vehicles, managers and investors determine the success of the portfolio’s companies exits. The study uses a database of PE/VC from FGVEAESP, collected in the First Brazilian Census of Private Equity and Venture Capital and the Guide GVcepe-Endeavor. The sample used in the analyses is composed of 63 investment vehicles that performed at least one exit, between 1999 and the end of the first semester of 2007. The percentage and the number of disinvestment made by initial public offerings (IPO), trade sale or secondary sale in the exits of the fund were used as measures of performance of the investment vehicles. According to the related literature, these mechanisms of disinvestment are the most efficient and profit means of liquidation of the private equity and venture capital participation in a portfolio company. The main findings indicate that only the investment vehicles and managers characteristics influence the success in the exits of PE/VC funds in Brazil, showing that the Smart Money doesn’t exist for the Brazilian PE/VC industry. Therefore, the evidences found lead to the conclusion that it is not possible to assure that some investors have the ability to identify the best managers, investing with them and than achieve better performance. The results of this study are robust and are consistent with the theory and the assumptions expected. Among the evidence found in the analysis, be highlighted that the following characteristics of the PE/VC organizations significantly affect the performance of investment vehicles: amount of committed capital, number of investments already made, the level of control of managers in the invested companies, the co-investments made, the presence of investors in investment committees, foreign origin of the PE/VC organization, focus on private equity firms, the percentage of selection of investments, intensity of contact between managers and companies in the portfolio and experience of the management team in the PE/VC industry. / Este trabalho investiga o efeito das características das organizações de private equity e venture capital (PE/VC) no desempenho dos respectivos fundos. Mais especificamente, a análise procura detectar como os perfis dos veículos de investimento, gestores e investidores determinam o sucesso nas saídas realizadas das empresas do portfolio. O estudo utiliza como base os dados de PE/VC da FGV-EAESP, coletados no Primeiro Censo Brasileiro de Private Equity e Venture Capital e no Guia GVcepe-Endeavor. A amostra usada nas análises é composta por 63 veículos de investimento que realizaram pelo menos uma saída, entre 1999 e o final do 1º semestre de 2007. Como medidas de desempenho dos veículos de investimento, foram utilizadas a porcentagem e o número de desinvestimentos realizados via abertura de capital (IPO), venda estratégica (trade sale) ou venda para outro investidor (secondary sale) das saídas realizadas pelo fundo. Segundo a literatura relacionada, estes mecanismos de desinvestimento constituem os meios mais eficientes e lucrativos de liquidação da participação de private equity e venture capital nas investidas. As principais conclusões encontradas indicam que apenas as características dos veículos de investimento e dos gestores influenciam o sucesso nas saídas dos fundos de PE/VC no Brasil, não se comprovando a existência do efeito Smart Money para esta indústria. Assim, as evidências encontradas levam à conclusão de que não é possível afirmar que alguns investidores possuem a habilidade de identificar melhores gestores, investir com eles e, assim, alcançar melhor desempenho. Os resultados obtidos neste estudo são robustos e estão de acordo com a teoria e com as hipóteses previstas. Dentre as evidências encontradas nas análises, pode-se destacar que as seguintes características das organizações de PE/VC influenciam significativamente o desempenho dos veículos de investimento: volume de capital comprometido, número de investimentos já realizados, nível de controle dos gestores das investidas, existência de co-investimentos, presença de investidores nos comitês de investimento, origem estrangeira da organização gestora, foco em empresas de private equity, percentual de seleção de investimentos, intensidade de contato entre gestores e empresas do portfolio e experiência da equipe gestora na indústria de PE/VC.

A commonly researched problem in computer science considers the different parameters of an evacuation simulation and how they affect evacuation times. Parameters like room size, obstacles and exit layouts have proven to have an impact on evacuation times. Previous research indicates that evacuation times are not always reduced when adding exits to virtual environments, suggesting diminishing returns of adding exits in crowd evacuation simulations. This raises the question if it is always advantageous to add exits with the goal of reducing evacuation times, and if the effectiveness of adding exits is dependent on factors like room size and layout. This thesis investigates the changes in evacuation times obtained by adding exits to virtual furnished environments using an agent- based orderly crowd simulator developed in Unity3D. The virtual environments used are modeled after three classrooms at The Royal Institute of Technology (KTH): a small classroom (E34), a medium sized classroom (D32) and a large classroom (Q33). The models are modified to accommodate one through six exits with the goal of determining if there are diminishing returns of adding exits in different crowd evacuation scenarios. The results indicate that diminishing returns in crowd evacuation times occur when adding exits to virtual rooms. Diminishing returns seem to appear at different points depending on the size and/or layout of the virtual rooms, appearing at higher exit amounts the larger the room is. Several limitations could have affected the results, which means that more research within the field is required before the findings are used in real world applications. / Ett ofta undersökt problem inom datavetenskap berör hur olika parametrar i en evakueringssimulering påverkar evakueringstider. Parametrar som rumstorlek, hinder och utgångslayout har visats ha en inverkan på evakueringstider. Tidigare forskning indikerar att evakueringstider inte alltid reduceras när man adderar utgångar till virtuella miljöer, vilket tyder på avtagande avkastning av att addera utgångar i evakueringssimuleringar. Detta väcker frågan ifall det alltid är fördelaktigt att addera utgångar i syfte att reducera evakueringstider, samt ifall effektiviteten av att addera utgångar beror på faktorer som rumstorlek och layout. Detta arbete undersöker förändringarna i evakueringstider som fås av att addera utgångar till virtuella möblerade miljöer genom att använda en agentbaserad ordnad simulator utvecklad i Unity3D. De virtuella miljöerna modelleras efter tre klassrum på KTH: ett litet klassrum (E34), ett medelstort klassrum (D32) och ett stort klassrum (Q33). Modellerna modifieras för att rymma en upp till sex utgångar med målet att bestämma ifall det finns avtagande avkastning med att addera utgångar i olika evakueringsscenarion. Resultaten indikerar att avtagande avkastning i evakueringstider förekommer när utgångar adderas till virtuella rum. Avtagande avkastning verkar framträda vid olika punkter beroende på de virtuella rummens storlek och/eller layout, och de framträder vid högre antal utgångar ju större rummet är. Flertalet begränsningar kan ha påverkat resultatet, vilket innebär att mer forskning inom ämnet behövs innan resultaten appliceras och används i verkligheten.