Deterioration models and market linked maintenance trigger based on big data for unpaved roads

Abstract

The consumer perception is a growing concern for the fresh farm produce market as external damage reduces the likelihood of purchase and thus reduces the economic value to the retailer and grower. One of the major causes of mechanical damage to the surface of the fruit results from the vertical vibrations generated by the movement of haulage trucks on roads with high surface roughness. Although careful handling and well-designed packaging reduce the loss of fruits due to bruising, this can only be altered to a certain extent beyond which it becomes economically unfeasible and environmentally harmful to increase the number of protective packaging layers (Steyn et al., 2015). Experiments conducted by Pretorius and Steyn (2012) have shown that although the length of an unpaved road section forming part of the transportation route from the grower to the marketplace may be 40 times shorter than the length of the paved road section, the transported freight accumulates more damage over the short unpaved section than over the significantly longer, smoother paved section. Thus the quality of unpaved roads has a significant effect on the profitability of commercial farming. The deterioration of roads is governed by the behaviour of the material and the capacity of the drainage system under the combined actions of traffic and the environment (Paterson, 1987). Effective pavement management is a challenging task due to increased pavement deterioration and performance demands along with limited budget and human resources. Effective maintenance can take place when decision making is informed by knowledge of the current condition and expected future condition of the road which can be derived by analysing the historical performance of similar road segment types. Extensive historical road roughness data has been collected using a response type road roughness measuring system over a period of 16 months for various unpaved roads in Limpopo, South Africa. This project used the historical road roughness data, supplemented by maintenance history and rainfall data, to generate roughness deterioration models for unpaved road sections with varying characteristics by fitting various mathematical functions to the collected field data. Simple annual maintenance cost models were derived from the roughness deterioration models and used in conjunction with previously developed tomato shelf-life models to obtain optimum roughness thresholds for road sections with varying characteristics and tomatoes with varying degrees of ripeness. The classification of ripening tomatoes has been formally standardised by the United States Department of Agriculture into green, pink and red stages of ripeness. In general, greater deterioration occurred over the dry season with the most rapidly deteriorating road section type having a slope of greater than 6 per cent as well as relatively heavy traffic. Exponential models were found to be the most appropriate to model deterioration of the road condition over the dry season while linear models were best suited for modelling deterioration over the wet season using the road roughness as a condition indicator. The optimum roughness thresholds were highest for pink tomatoes and lowest for red tomatoes thus indicating that the lowest optimal maintenance expenditure is achieved when transporting red tomatoes while foregoing an extended shelf-life. The highest optimal shelf-life is reached when transporting pink tomatoes while incurring a greater maintenance cost. Thus, the results indicate that the optimum scheduling of maintenance activities on unpaved roads is dependent upon the demand of the product at the market. Greater demand for tomatoes will require a lower optimal shelf-life, thus the maintenance cost is minimised by transporting red tomatoes and scheduling maintenance activities using roughness triggers ranging between 4.4 m/km and 5.6 m/km (which is dependent upon the slope and traffic on the road section). Lower market demand will require a higher optimal shelf-life, thus the maintenance cost is minimised by transporting pink tomatoes while scheduling maintenance activities using roughness triggers ranging between 3.9 m/km and 4.4 m/km. The difference between these two scenarios results in an average shelf-life increase of 48 per cent while increasing the maintenance cost by an average of 82 per cent.