Using artificial neural network and multivariate analysis techniques to evaluate road operating conditions

Regional paved roads are low volume roads with a prevalence of heavy traffic. In the world, these roads concern about 80% of the total road network; however, the traffic that affects these roads is about 20%. Since regional roads are characterized by weak demand, budget for their management/maintenance is very low. This produces considerable difficulties in the choice of strategies for maintenance planning and scheduling. For this reason, the recurring topics of research in this field deal with typical roads issues and aim to develop low cost tools and methods. The study proposes a decision support system to evaluate regional paved roads operating condition in relation to the hydrogeological situation. In particular, the system allows to evaluate in a quick and easy manner, the operating conditions of the road, through low-cost tools (i.e. using low economic resources). This is very useful in the case of LVRs because administrations for these roads have a limited budget. The procedure is developed on a regional paved roads network based on more than 80 roads located in Southern Italy. Data is collected by direct surveys in the field and is integrated with cartography and information available in road agency records. From data analysis, obtained using two different techniques, an easy and quick use procedure is made. In particular, Model 1 is built through multivariate analysis and Model 2 using the artificial neural network (ANN) technique. The results show the validity of the two models in Regional paved roads operating conditions estimation in relation to hydrogeological situations of sites. Both models show good reliability. In particular, the first model (Model 1) is characterized by a high level of significance (p < 0.01) and by a coefficient of determination equal to 0.82. Comparative tests between the second model (Model 2) on which standard tests cannot be performed for obvious reasons, and the first model (Model 1). The results show that the ANN model (model 2), characterized by lower residual, simulates more accurately than the second (Model 1).     

How Much Will It Cost to Join the Club? The Extra Costs of Approximating Lithuanian Environmental Laws with Those of the European Union

This paper presents estimates of costsassociated with approximating Lithuanianenvironmental protection legislation with thatof the European Union (EU). Lithuania is oneof twelve EU associate members in Central andEastern Europe that is currently preparing foraccession by approximating their legislationwith that of the EU. The costs of fifteen EUdirectives are considered and details are givenon four directives. Necessary investments maytotal Euro 1500 million by 2015 and the presentvalue of all annualized costs is expected to beabout Euro 2200 million. In 2015, annualized costsare estimated to be approximately Euro 500 million,but these costs are only part of the totalcosts of approximation. Non-environmentalcosts are not considered. Assuming a modestlyambitious average annual growth of GDP of 2.0%per year implies that approximation with thefifteen directives analyzed will cost roughly3.5% of GDP in 2015. This level of additional commitment to environmentalprotection is itself much higher than the 2.0%of GDP being spent on average by OECD countriesand suggests the possibility of a substantialeconomic burden on the Lithuanian economy. Public budgets and households are expected tocarry a substantial portion of this cost,because many directives are the responsibilityof national and local governments. Making theright choices that are expected to be part ofapproximation with the environmental acquis is likely to benefit from carefulcomparisons of costs and willingness to pay forthe environmental benefits of approximation.