Risk management during planning and construction of large infrastructure projects for improving their sustainability

[M.S. Mikić]

439 listova

Građevinarstvo / Civil Engineering Menadžment i tehnologija građenja, Ekološki inženjering / Building Management and Technology, Environmental Engineering

Investicioni projekat u građevinarstvu se definiše kao kompleksan tehničko-tehnološki, organizacioni, pravni, ekonomski i finansijski poduhvat koji se sastoji od skupa koordinisanih i kontrolisanih aktivnosti sa jasno definisanim početkom i krajem, čiji je cilj izgradnja, rekonstrukcija, modifikacija i/ili opremanje objekta ili objekata koji su potrebni investitoru. Kompleksnost je, kao karakteristika projekta i okruženja u kome se projekat realizuje, naročito izražena kod velikih infrastrukturnih investicionih projekata. Predmet istraživanja u ovoj disertaciji jesu kapitalni infrastrukturni projekti velike investicione vrednosti, veće od pedeset miliona evra. Razmatrani su projekti izgradnje objekata sistema ekonomske materijalne infrastrukture, i to: objekata saobraćajne (autoputevi, železničke pruge, metro linije i aerodromi), energetske (objekti za proizvodnju i prenos električne energije i gasa) i hidrotehničke (hidrotehničke konstrukcije) infrastrukture. Ovi projekti predstavljaju motore razvoja svakog društva i države. S obzirom na potencijalne dugoročne ekonomske, društvene i ekološke efekte koje veliki infrastrukturni projekti mogu proizvesti, proces njihovog pokretanja, planiranja i realizacije je uvek praćen značajnom pažnjom celokupnog društva. Prethodna istraživanja pokazuju da je i pored značaja koji kapitalni infrastrukturni projekti imaju, njihova realizacija veoma često neuspešna, i to kako u pogledu neispunjenja tradicionalnih kriterijuma uspeha projekata (troškovi, vreme, kvalitet), tako i u pogledu negativih ekonomskih efekata projekata i negativnih efekata projekata na društveno i ekološko okruženje. Ova tri aspekta uticaja na okruženja projekta (ekonomski, društveni i ekološki uticaji) čine tri aspekta održivosti i primene principa održivog razvoja na konrektnom projektu. Kao glavni razlozi za odstupanja od planiranih performansi realizovanih kapitalnih infrastrukturnih projekata u literaturi se navode rizici koji proističu iz specifičnosti ovakvih projekata. Kapitalni infrastrukturni projekti su, u odnosu na građevinske projekte manje investicione vrednosti, investiciono zahtevniji, značajno kompleksniji, prisutna je veća neizvesnost, veći broj učesnika, dugotrajniji su, i njihovi potencijalni efekti na ekonomiju, društvo i okolinu su veći, dalekosežniji i privlače više pažnje javnosti. Prethodna istraživanja takođe otkrivaju da, za sada, ne postoji jedinstvena metodologija za procenu održivosti infrastrukturnih objekata u ranim fazama realizacije investicionog projekta. Konačno, oblast upravljanja rizicima prilikom realizacije infrastrukturnih projekata nije u dovoljnoj meri obrađena u domaćoj stručnoj i naučnoj literaturi. U prvom delu istraživanja za potrebe ove disertacije sprovedeno je kvantitativno istraživanje sa ciljem identifikacije ključnih rizika po ostvarenje troškovnih i vremenskih performansi projekata izgradnje infrastrukturnih objekata u Srbiji. Osim toga, ispitano je i postojanje prakse upravljanja rizicima na investicionim projektima u Srbiji. Dobijeni rezultati tog dela istraživanja su pokazali da su najznačajniji rizici pri realizaciji infrastrukturnih objekata u Srbiji: Nedostatak finansijskih sredstava za realizaciju projekta, Finansijski rizik, Politički rizik i Korupcija. Takođe, bez obzira što postoji svest da je primena upravljanja rizicima važna i što postoji potreba za njegovom primenom, u Srbiji se upravljanje rizicima ne primenjuje dovoljno i prisutan je manjak znanja u vezi sa upravljanjem rizicima. Međutim, među praktičarima iz oblasti građevinarstva postoji izražena zainteresovanost da se o upravljanju rizicima nauči više. S obzirom na ovakve rezultate, u daljem istraživanju analizirani su i prikazani mogući pristupi za analizu rizika i integraciju analize ekonomske, socijalne i ekološke održivosti projekta u fazi formiranja koncepcije kapitalnog infrastrukturnog projekta. Predložena su i razmotrena dva pristupa – kvalitativna i kvantitativna analiza rizika, uz primenu socijalne Cost-Benefit analize (SCBA). Mogućnost i implikacije praktične primene predloženih pristupa analizirani su na primeru dve studije slučaja kapitalnih infrastrukturnih objekata (izgradnja postrojenja za insineraciju komunalnog čvrstog otpada i izgradnja deonice auto-puta). U disertaciji je pokazano da primena predstavljene SCBA uz monetarizaciju eksternih efekata projekata omogućava svođenje na istu meru i sveobuhvatno sagledavanje potencijalnih ekonomskih, društvenih i ekoloških uticaja projekta kroz ceo životni ciklus. Primena kvalitativne analize rizika omogućava pravovremenu identifikaciju, rangiranje i sistematičan prikaz potencijalnih pretnji po ostvarenje pojedinih ciljeva projekta, te predlog mera za otklanjanje ili umanjenje pretnji. Stohastički pristup i Monte-Carlo analiza za kvantitativnu analizu rizika u studiji opravdanosti doprinose većoj pouzdanosti procene finansijskih i društveno-ekonomskih rezultata projekta. Prikazanu metodologiju i pristup je moguće koristiti u budućim predinvesticionim analizama kapitalnih infrastrukturnih objekata, naročito na domaćem tržištu i tržištu zemalja u razvoju. Cilj daljeg istraživanja bilo je ispitivanje mogućnosti za razvoj ekspertskog sistema za procenu rizika u ranim fazama realizacije kapitalnih infrastrukturnih projekata. Izvršena je provera hipoteze da je na osnovu poznatih istorijskih podataka o ostvarenju planiranih troškova i rokova realizacije i o karakteristikama realizovaniih kapitalniih infrastrukturniih projekata i njihovog okruženja moguće napraviti model za predviđanje uspešnosti realizacije na novim projektima, ukoliko su karakteristike novih projekata i njihovog okruženja poznate. Usvojena metodologija istraživanja podrazumevala je najpre prikupljanje podataka sa realizovanih kapitalnih infrastrukturnih projekata, njihovu pripremu a zatim analizu primenom metoda mašinskog učenja. Mašinsko učenje je oblast kompjuterskih nauka koja se bavi kreiranjem i analizom metoda na kojima počivaju računarski programi koji uče iz iskustva. Prikupljeni su podaci o 30 saobraćajnih, 12 energetskih i 2 hidrotehnička kapitalna projekta (ukupno 44 projekta, svaki vrednosti preko petsto miliona evra) realizovana na teritoriji Evrope. Podaci su sistematizovani u obliku 3 binarna pokazatelja uspešnosti projekata (prekoračenje troškova, kašnjenje u fazi građenja i kašnjenje u fazi planiranja) i 46 binarnih atributa projekata, koji opisuju izvore rizika iz 5 kategorija: učesnici na projektu (interni i eksterni), spoljašnje okruženje projekta (pravno, društvenoekonomsko, političko), upravljanje projektom, tehnološki aspekti, razno. Metodologija prikupljanja i pripreme podataka zasnivala se na metodologiji rada formiranoj u okviru međunarodne naučne COST akcije TU1003: "Megaproject – Efficient Design and Delivery of Megaprojects in the European Union". Zatim je, prema originalno osmišljenoj metodologiji, zasnovanoj na prethodnim istraživanjima u oblastima procene rizika i primene metoda mašinskog učenja u upravljanju projektima, izvršena uporedna analiza performansi dvanaest predloženih modela za predviđanje prekoračenja Miljan S. Mikić, dipl. građ. inž. 8 planiranih troškova izgradnje, kašnjenja u fazi građenja, kao i kašnjenja u fazi planiranja realizacije projekata. Ispitana je mogućnost kombinovane primene statističkih metoda (Selekcija podskupa atributa zasnovana na korelaciji i Selekcija zasnovana na vrednosti informacionog dobitka) i metoda mašinskog učenja (Metoda vektora podrške, Veštačke neuralne mreže, K-najbližih suseda, Drvo odlučivanja, Naivni Bajesov klasifikator i Logistička regresija). Dobijeni rezultati su dokazali da je za dati skup prikupljenih podataka bilo moguće napraviti modele za predviđanje navedenih pokazatelja uspešnosti u ranoj fazi realizacije kapitalnih infrastrukturnih projekata. Istraživanje predmetnog skupa podataka je takođe identifikikovalo podskupove od relativno malog broja ključnih izvora rizika iz faze pre početka građenja (3-10, zavisno od problema) čije poznavanje je dovoljno da se ostvare dobijene, relativno visoke performanse predviđanja. Za dati skup prikupljenih podataka, najznačajnije identifikovane kategorije rizika jesu: Društveno-ekonomsko okruženje projekta, Eksterni učesnici na projektu i Tehnološki aspekti projekta. Veoma bitna distinkcija primenjenog pristupa u odnosu na analizu korelacije pojedinačnih izvora rizika i pokazatelja performansi projekata jeste da se ovde nastoji utvrditi zajednički uticaj koji više izvora rizika istovremeno imaju na performanse projekata. Primena predloženih modela, za rano predviđanje uspešnosti realizacije projekata, najveću korist donela bi investitoru i donosiocima odluka u ranoj fazi realizacije projekta, jer bi mogla da pruži bolji uvid u očekivane performanse datog projekta, kao i da upozori na izvore rizika zbog kojih bi performanse projekta potencijalno mogle biti ugrožene. Kako bi ekspertski sistem za procenu rizika prilikom pokretanja i realizacije kapitalnih infrastrukturnih projekata bio zaokružen, neophodan je dalji rad na dopunjavanju baze podataka.

A construction investment project is defined as a complex technical and technological, organizational, legal, economic and financial endeavour that consists of a set of coordinated and controlled activities with a clearly defined beginning and end, with the goal of building, reconstructing, modifying and/or equipping a facility or facilities that are required by the investor. Project and project environment complexity is particularly emphasised on large infrastructure investment projects. The subject of the research in this dissertation are large infrastructure projects with an investment value of more than fifty million euros. The projects of planning and construction of hard (physical) economic infrastructure were investigated; those included ones in the traffic (highways, railways, subway lines and airports), energy (facilities for the production and transmission of electricity and gas) and hydro-technical sector. The above listed projects represent the development engines of any society and state. Given the potential long-term economic, social and environmental effects that large infrastructure projects can produce, their planning and construction process is always followed closely by the entire society. Previous research shows that, despite the importance of large infrastructure projects, their implementation is often unsuccessful, both in terms of failure to meet the traditional project success criteria (cost, time, quality), as well as in terms of the adverse economic, social and environmental effects that the projects can create. Analysis of these three impacts (economic, social and environmental), as aspects of sustainability, allows for the incorporation of sustainable development principles within a specific project. In the literature, risks that arrise from specific characteristics of large infrastructure projects are identified as the main cause of deviations from the planned performance of large infrastructure projects. In comparison to construction projects of smaller investment value, large infrastructure projects are: financially more demanding; significantly more complex; carry a greater uncertainty; include more stakeholders; are Miljan S. Mikić, dipl. građ. inž. 11 longer lasting, and their potential effects on the economy, society and the environment are greater, more far-reaching and generate more public attention. At present, there is no single accepted methodology for sustainability assessment in the early phase (phase of conducting pre-feasiblity anf feasibility study) of an investment project. Additionally, the area of risk management on infrastructure projects is not sufficiently addressed in the domestic professional and scientific literature. In the first part of the dissertation, quantitative research was conducted in order to identify the key risks to achieving cost and time performance of infrastructure construction projects in Serbia. In addition, risk management practices related to investment projects in Serbia were investigated. Research results showed that the most significant risks associated with construction infrastructure projects in Serbia are: the lack of funds for project implementation; financial and political risks; and corruption. Also, regardless of the fact that there is awareness of the importance of risk management and the need for its implementation, risk management is not implemented well in Serbia and there is a lack of knowledge in relation to risk management. However, there is a strong interest to learn more about risk management among the practitioners in the construction field. Considering obtained results, in further research, possible approaches for risk analysis and integration of project economic, social and environmental sustainability analysis were addressed. Two approaches were proposed and investigated – a qualitative and a quantitative risk analysis, along with applying a Social Cost-Benefit Analysis (SCBA). The possibility and implications of the practical application of the proposed approaches were analysed on two case studies of major infrastructure projects (Municipal solid waste incineration plant and a Motorway section). The application of the presented SCBA with monetization of the project external effects allows for comprehensive consideration of the potential economic, social and environmental impacts of the project throughout the entire life cycle of the infrastructure facility. The application of qualitative risk analysis enables the timely identification, ranking and systematic overview of potential threats for the achievement of specific project objectives and the proposal of measures for the elimination or reduction of threats. The stochastic approach and Monte-Carlo Analysis for quantitative risk analysis in the project feasibility study allows the higher reliability of assessment of the project financial and socio-economic results. The studied methodology and approach can be used in future pre-investment analyses of major infrastructure facilities, especially in the domestic market and the market of developing countries. The further objective of this dissertation was to examine the possiblity for developing an expert system for risk assessment in the early phase of large infrastructure projects. The hypothesis was that based on the known historical data on the achievement of planned cost and deadlines and the characteristics of large infrastructure projects and their environment, it is possible to create a model that can predict the success or failure (regarding cost and time performance) of new projects, if the characteristics of new projects and their environment are known. The adopted research methodology consisted of data collection from completed large infrastructure projects, data preparation and analysis using machine learning methods. Machine learning is a field of computer science that deals with the creation and analysis of methods that are used by computer programs that learn from experience. In previous studies, it has been proven that certain machine learning methods can be successfully used to predict the performance of construction investment projects. Data from 30 traffic, 12 energy and 2 hydro-technical large infrastructure projects (a total of 44 projects, each with the investment value of more than five hundred million euros) completed in Europe were collected. The data were transformed to the form of three binary project success indicators (Cost overrun, Delay in the construction phase and a Delay in the planning phase) and 46 binary project attributes, which describe the sources of risk from five categories of project characteristics and characteristic of the project environment: Project stakeholders (internal and external); The external environment of the project (legal, socio-economic, political); Project management; Technological aspects; and Miscellaneous. The methodology of collecting and preparing the data was based on the methodology of work of the international scientific COST Action TU1003: "Megaproject - Efficient Design and Delivery of Megaprojects in the European Union". Then, according to the newly proposed methodology that was based on previous research in the areas of risk assessment and the application of machine learning methods in project management, a comparative analysis of the performance of the twelve proposed models for the prediction of the exceedance of the planned construction costs, the delay in the construction phase, as well as the delay in the planning phase execution of the projects was performed. The possibility of the combined application of statistical methods (the selection of a subset of attributes based on correlation and the selection based on the value of information gain) and machine learning methods (Support vectors machine, Artificial neural networks, K-nearest neighbour, Decision tree, Naive Bayesian classifier, Logistic regression) was analysed. The results have shown that, for a given set of collected data, it was possible to build models that predict success indicators in the early implementation stage of large infrastructure projects. The research also resulted with the identification of subsets of a relatively small number of key sources of risk from the pre-construction phase of project development (3-10 sources of risk, depending on the problem) whose knowledge is sufficient to yield relatively high performance predictions for a given set of collected data. The most important identified risk categories are: the socio-economic environment of the project; the external stakeholders in the project; and the technological aspects of the project. A very important distinction between the approach applied in this dissertation and the correlation analysis of individual sources of risk and project performance indicators is that in this dissertation the attempt was to determine the combined concurrent effect of multiple sources of risk on project performance. The greatest benefit of the application of proposed models for the early prediction of project success is to the investor and the decision-makers at the early stage of a project, as the models can provide a better insight into the expected performance of a given project, as well as to draw attention to the sources of risk that would potentially endanger project performance. In order to finalize the expert system for risk assessment during the planning and construction phases of large infrastructure projects, further research should be aimed at broadening the database of large infrastructure projects.

Upravljanje projektima, upravljanje rizicima, procena rizika, infrastruktura, kapitalni projekti, mašinsko učenje, održivost

Project Management, Risk Management, Risk Assessment, Infrastructure, Large projects, Machine Learning, Sustainability

69.05:005.8]:502/504(043.3)

Tekst