|dc.description.abstract||Road transport is an important sector of economic activity, especially in developing countries, where it plays an essential role in marketing agricultural products and providing access to health, education and agricultural inputs and extension services.
Zambia as a land locked country has been working hard to become “land linked”. The country has the potential of attracting more transit traffic and becoming a regional distribution centre for all kinds of goods and commodities. The Zambian road network is one of the country’s largest public sector assets. It is therefore essential that this vital asset is managed efficiently and effectively, invariably within a constrained budgetary situation, in support of socio‐economic growth and the development of the country. It is perceived by stakeholders that Zambia does not always get value for money in road infrastructure delivery. Industry regulators and public institutions have indicated that there was a notable trend in varying costs of construction from project to project and from one public institution to another, that it had become increasingly difficult to ascertain the true cost of projects and thereby unable to guarantee value for money. The research aimed at developing a unit cost estimation model (UCEM) for roadworks incorporating neural network (NN) to provide a standardised procedure of pricing road activities in Zambia and help understand prevailing market rates in the Zambian road sector (ZRS). The research involved establishing the base rates and determining the economic strata to form the unit rates used in the model. The base rates of the 854 pay
items from the Southern Africa Transport and Communications Commission (SATCC) Standard Specifications for Road and Bridge Works used in the ZRS were calculated from first principles. The economic strata involved establishing the cost factors that affect construction unit rates (CUR) in the ZRS. From reviewed literature, forty five (45) cost factors were identified. Expert opinion reduced the factors to thirty one (31). The expert opinion was obtained through the Delphi technique and Pareto analysis was used to further analyse the factors and reduce them to twenty five (25). Further information on the 25 factors was obtained through questionnaire survey. Using factor analysis the 25 factors were further reduced to eight (8). The established 8 factors namely: contractor capacity; project location; period of honouring payments; level of design; cost escalation; materials availability; country corruption profile; and political environment were identified as those that impact unit costs in the ZRS. The 8 factors were then v analysed using NN to determine the proportionate breakdown of the cost factors in a given unit rate. The UCEM incorporated quantitative base unit rates and quantified qualitative economic strata to establish the prevailing rate. The UCEM was validated using real system measurements from three (3) SADC countries and five (5) local road projects. The value of the study is to provide a standardised procedure of pricing road activities to ensure uniformity in public and private procurement practice in the ZRS. The UCEM would assist those involved with roads project estimating to calculate the Engineers’ estimate with a fairly high level of accuracy. Finally, it is hoped that the model would provide a generic acceptable rate analysis system that can be used as a basis to compare against future projects. Key words: construction unit rates, neural networks, cost factors, unit cost estimation model, Zambian road sector.||en