Capacity Planning in a Sugarcane Harvesting and Transport System Using Simulation Modelling

REDUCING costs within the harvesting and transport system is a high priority for many sugar milling regions in Australia. Typical issues include reducing the number harvesting groups, harvesting over a longer time window in a day, rationalising infrastructure and achieving a better co-ordination between harvesting and transport activities. As part of a series of integrated models to conduct the analysis, a simulation model for capacity planning was developed to estimate the: 1) number of locomotives and shifts required; 2) the number of bins required; and 3) the period of time harvester operators spend waiting for bins. The model belongs to the category of planning models for unscheduled traffic, which means a locomotive schedule does not need to be produced. While new for the Australian sugar industry, these types of models have been used in the past extensively for planning in road and urban/freight railroad systems. Some key advantages of the model versus a scheduling tool are the ability to: 1) work in situations with high probabilities of delay and down time in the transport system; 2) measure capacity independently of a schedule or in applications where it is impossible to produce an effective transport schedule using a model; and 3) fully integrate with other models within the sugar harvesting and transport system for whole-of-system optimisation. The benefits of the model are demonstrated through application to the Mourilyan case study, to provide the region with an understanding of the impacts from: 1) removing double handling of bins; 2) extending the time window of harvesting; 3) reducing the number of harvesting groups; 4) and upgrading bin fleet and sidings. A scenario for harvesting during a time window of 18 hours was piloted within Mourilyan during the 2003 harvest season. The benefits of integrating the capacity planning model with a model for scheduling harvesters into sidings is demonstrated with the Mossman case study, showing significant reductions in the daily variability of demand on the transport system. Introduction Several mill regions within the Australian sugar industry are exploring opportunities to reduce costs within their harvesting and transport system. Scenarios include extending the time window of harvest, amalgamating harvesting groups, rationalising/upgrading transport infrastructure, implementing harvest best practice and removing some of the inefficient practices in cane transport such as the double handling of rail bins. Higgins, A. and Davies, I. Proc. Aust. Soc. Sugar Cane Technol., Vol. 26, 2004 _________________________________________________________________________________________________ Addressing most of these required the systems modelling of harvesting and transport combined since they are closely linked activities. A modelling framework of the harvesting and transport system was developed (Higgins et al., 2003) and is illustrated in Figure 1. It shows some of the major interactions between different models for transportation and harvesting planning. The Harvest Haul model (Sandell and Prestwidge, 2004) and harvester/siding rostering models (Higgins and Postma, 2003) of Figure 1 are existing models redeveloped for a whole-of-system modelling capability. Other models developed for the framework were a model for optimising the location of harvesting groups and rail sidings/loading pads, a financial model for calculating the costs and benefits to the industry sectors (Antony et al., 2003), and a simulation model for capacity planning in harvesting and transport. The objective of this paper is to highlight the capacity planning model and its benefits using case studies in Mourilyan and Mossman. Row length Bin and locomotive fleet Harvesting information Harvest Haul Model Capacity Planning Model Harvest Group and Siding Location Model Number of groups and hours of harvest Haulout distance Locomotives, shifts and bins required Number and location of sidings Harvesting costs Financial Model Waiting time for bins Siding and locomotive upgrade Upfront and on-going costs of harvesting and transport Harvester and Siding Rosters Location of harvesters Fig. 1—The harvest-transport modelling framework. Capacity planning in transportation is a large research field in the literature for which most models fall into two major categories: 1) models based on scheduled traffic; and 2) models based on unscheduled traffic. While cane transport models are not new to the Australian sugar industry (Pinkney and Everitt, 1997; Grimley and Horton, 1997), they are based on scheduled traffic. This means each scenario requires the production of a locomotive or truck schedule in order to assess regional characteristics such as and locomotive and bin requirements. The capacity planning model of this paper is based on unscheduled traffic which means the outputs (locomotive and bin requirements, waiting time for bins, etc.) are calculated without first producing a traffic schedule. While new for the sugar industry, capacity planning models based on unscheduled traffic are commonly used for freight railway applications (Higgins and Kozan, 1997) for urban planning; and Chen and Harker (1990) for single-line freight planning). Models based on producing schedules have advantages over models based on unscheduled traffic in that they provide schedules for operational use, which in turn allows a traffic officer to immediately see how it would work in practice. Higgins, A. and Davies, I. Proc. Aust. Soc. Sugar Cane Technol., Vol. 26, 2004 _________________________________________________________________________________________________ However, models based on unscheduled traffic have the following advantages over those that produce schedules: • They better handle scenarios where it is difficult/impractical to produce a schedule. Some strategic scenarios, such as redesigning a transport track system for use in five years time, have a large amount of unknowns at the operational level needed for an effective application of a scheduling tool. It can also be impractical to apply a scheduling model when a schedule requires several major changes during the day. • They cater for a transport system that has large amounts of unforeseen delay (e.g. locomotive breakdowns), by incorporating probability distributions to account for such delays. Models based on producing schedules only capture certainty and do not capture the unforeseen events that lead to increased infrastructure requirements and waiting time for bins. • They produce scenarios independent of a schedule. When using a model based on scheduled traffic, the outputs of locomotive and bin requirements are biased towards the schedule it was based on. That is, if the schedule is changed, so are the outputs. This makes it difficult to make fair comparisons between scenarios if their schedules are very different. • Models based on scheduled traffic require large resources to apply compared to models based on unscheduled traffic, since more detail is required at the operational level to produce schedules. This is an important characteristic because some scenarios require assessing impact across each day of the harvest season, while a scheduling model would require a large amount of resources to set-up. • It is very difficult to explicitly integrate models based on scheduled traffic with other models in the sugar value chain for whole-of-system analysis and optimisation, compared to models based on unscheduled traffic. For example, using a scheduling tool, it is almost computationally impossible to simultaneously optimise the start time combinations of harvesters and their migration schedule across sidings to achieve the most efficient utilisation of the transport system. The capacity planning model calculates the following key attributes across the harvest season: number of locomotive shifts and bins required; and the time harvesters spend waiting for bins when there are limitations on shifts and bins. Other attributes such as cut-to-crush time, siding and yard utilisations can also be calculated. In this paper, we highlight the simulation model for capacity planning in harvesting and transport, along with the benefits of the model through applications to the Mourilyan and Mossman case studies. The latter case study highlights the benefits of the model within a whole-of-system optimisation framework through integrated rostering of harvesters into sidings and transport planning, which is a modelling capability not available to the sugar industry in the past.