Increasing fuel costs as well as changes in service hours and capacity have significantly reduced profits for both transport companies and their customers.
However, logistics planning is often palmed off on a couple of people beavering away in a back room with pen, paper and an Excel spreadsheet. Given their low status and lack of up-to-date tools it is not surprising that the planners’ suggestions for operational improvements frequently receive short shrift from the board.
But this approach, which worked fine when the world was simpler and time less pressing, can’t hack it now. Modelling the alternatives involved in making 100 deliveries with a maximum of five stops can produce over 79 million solutions for a third-party logistics company that wants to meet set delivery dates with the fewest trucks possible and at the lowest cost.
Trying to plan manually or modeling warehouse, yard and transport operations separately has led to yawning gaps between the models and reality.
Increasingly complex distribution networks, pressure to reduce inventory and operate according to just-in-time principles mean the mathematics involved in optimisation is getting harder. Add in constraints of capacity, security and flexible shift patterns and arriving at the best solution is a major headache. The modeling process is often difficult and requires costly consultants to tailor a model.
It doesn’t have to be this way. Techniques for building computer models that find optimal solutions to logistics problems and enable companies to compare one plan with another have been around for some time, but until recently they tended to be the preserve of large organisations with critical transport problems such as train companies and airlines.
These days, thanks to more powerful computing and advances in the algorithms, or the problemsolving elements of programs, companies are applying optimisation software to a far greater range of logistics issues. Over the past 10 years solution times have improved 100-fold and problems considered out of reach only a few years ago can now be solved – often in real time.
There are two techniques for working out optimisation problems. Mathematical programming approaches are best for planning applications. They are good at finding optimal solutions, computing them rapidly with algorithms. Mathematical programming technologies have been successfully employed in the manufacturing, transport, telecommunications, natural resource and utility industries since the 1960s.
More recent constraint programming techniques are better suited for operational problems such as scheduling, sequencing, configuration and routing – problems that need to represented with logical expressions and that require fast solutions. Constraint programming uses information contained in the problem to prune the search space. Users can guide the search process with their own knowledge of the problem.
Often logistics tasks such as scheduling possess both planning and operational aspects and call for a hybrid approach involving mathematical and constraint programming.
Improvements in manipulating and displaying the output from optimisation programs have also contributed a lot to their usefulness. For example, Deutsche Bahn, Germany’s national railway, recently used graphics displays on a new train dispatching system to give dispatchers a total picture of scheduling information. Colour coding of trains and context sensitive dialogue boxes have made scheduling easier and cut training time for new employees.
For all the advances in IT, many firms still struggle with the technology. ‘The logistics industry has built a lot of models but if you look at the market only 20 per cent of firms are using them effectively,’ says Christophe Gasc, UK managing director of software company ILOG. ‘Most still don’t have accurate planning and have gone back to Excel sheets, or running planning as a separate activity from running the company.’
ILOG has recently introduced a transport planning application.
Today, reductions in inventory mean there must be better collaboration between transport and production staff, as well as the analytical and operational tools they use.