INTELLIGENCE IN THE WAREHOUSE

To increase output of distribution centers as well as to reduce operating costs it is not always necessary to physically restructure the warehouse layout or to invest in expensive warehouse technology. In fact, a more intelligent handling with existing assets is often sufficient. As examples like Avon Cosmetics or BMW demonstrate, intelligent software tools in addition to traditional WMS help to identify and realize appropriate measures.

Nowadays, many companies look to optimize their distribution processes to keep their competitiveness and to meet market requirements. Particularly against the background of the global financial crisis, managers are searching for ways to improve their cash flow – in internal logistics as well.

At the same time, while service quality levels need to be maintained, a cost reduction environment makes huge investments difficult. The use of intelligent software, then, can help to use existing resources in a more efficient way and realize various types of optimization potential as detailed in Table 1.

SMART PLANNING

A good example for intelligent logistics optimization can be found at Avon Cosmetics. As the world’s largest direct to consumer seller with US$10 billion in annual revenue, Avon markets to women in more than 100 countries through 5.4 million independent sales representatives.

For Avon’s shipping department, which is responsible for the on-time distribution of all ordered items at a minimum cost, its most important concern is the optimal assignment of the products and the equal workload of the different stations of the warehouse. During this process numerous constraints regarding the positioning of the products have to be observed, such as:

• Bulky products must be assigned to big bins

• Heavy products must be assigned to bottom bins

• Product look-alikes should not be assigned next to each other

• Valuable items must be assigned to “gold” stations

• Sales brochures must be assigned to special bins or in special stations

Before starting each sales campaign, Avon obtains new estimates from the marketing department. But even using sophisticated tools, there are variations between estimates and actual sales – and any bottleneck in the system creates a very expensive problem. For example, if there are 10 stations in a line and one station is overloaded by 10 percent, the result will be one station working at 100 percent and nine stations working at 90 percent. The average is a workload of 91 percent, meaning that nine percent of productivity is lost!

Another dilemma is the filling up of cartons, which are then shipped to the customer. If more cartons are needed, this in turn increases the shipping expense. On the other hand, if more cartons are used, then the filling up of the cartons is easier and faster.

Avon adopted an intelligent system to manage its shipping floor, one which uses a combination of traditional operations research algorithms, such as MILP (mixed integer and linear programming), and new genetic algorithms that imitate nature’s drive for a better fit regarding existing situations and search algorithms. The optimization software integrates frequent moves, optimal picking lists, balanced workloads, best carton fi lling and smooth material flows.

At the Avon branch in Montreal, Canada, during four campaigns, the new system ran parallel to the old system and the following results were reported:

• Planning time was cut by more than 50 percent.

• Due to this increase in productivity, four stations could be closed

• The workload for each station was better balanced, enabling a smoother flow of cartons through the shipping floor.

The software provides three key functions, as detailed below.

1. Shipping line balance

On the shop floor, products have to be assigned to positions where they will be picked or dispensed from. Using this module, the software assigns the products with the highest estimates to the best locations. The shipping floor layout is modeled once, considering various location properties such as dimensions, security provisions and other constraints. Differences between good and bad picking locations are expressed in terms of costs. Four different kinds of costs are considered:

• Picking costs – arise when a product is picked from the bin into the carton, and these costs can be measured for every bin either in terms of time or distance. From the picking costs, the workload of each station can be derived.

• Changeover costs – define the expenses for moving a product from one bin to another. Again, distances and times are transformed into costs. If, for example, after the end of a campaign the estimates for a product for the following campaign are low, the optimizer could suggest moving that product to a “worse bin” (e.g. a bin with higher picking costs).

• Replenishment expenses – define how much it costs to refill a bin with another box of the same product during a campaign.

• Penalty costs – provide a measure of the “ideal” assignment, such as heavy products to bottom bins. Thereby only assignments to less favoured bin locations will be penalised.

The costs are defi ned once during the implementation phase, and the product information and estimates are loaded. On pressing a button, the system creates the lowest possible cost assignment of products to bins, considering all constraints. Using genetic algorithms, the workload is then evenly distributed over all stations in order to avoid bottlenecks. If for any reason manual assignments for some products are preferred, the products can be moved interactively in a drag-anddrop editor.

2. Cartoning

The software’s cartoning algorithm selects items from the orders and fills them to the optimum size and number of cartons, ensuring throughput while maximising productivity, quality, and minimizing product display space. A minimum number of cartons for all orders has to be achieved.

Due to differing layouts between branches, flexible solutions have to be provided. Cartons may start at multiple start points or travel through mirrored or non-mirrored environments. The content of the cartons can be user defined.

Cost savings can be achieved in various ways. For example, cartons of the same order run independently through different shipping floors. Only required cartons enter a station and full cartons immediately leave the shipping floor via express lines. Therefore, everything results in a more balanced workflow.

3. Flow optimization

The most sophisticated part of the new system where the model of the shipping floor for the modules shipping line balance and cartoning is enlarged and contains the complete material flows. if a day’s volume consists of 10,000 orders, then approximately one billion activities must be simulated, which takes only a few minutes on a PC.

ALL IN ORDER

Another case study for intelligent logistics optimization can be found at the BMW plant in Dingolfi ng, Germany, which is the largest manufacturing site within the worldwide BMW production network and forms the centre of global spare parts supply. More than 200 trucks, more than 40 containers and 300 m3 of air cargo supplied with components depart from Dingolfi ng every day.

Using order sequencing software to process its spare parts orders allows the optimal bundling of incoming orders to batches and minimizes picking efforts. Generating respective picking instructions, the software ensures that every picking bin passes as few storage areas as possible, taking into account specifi c restrictions such as express orders or product properties.

Since the introduction of the optimization solution the BMW spare parts center benefi ts from shorter picking routes, reduced costs and a more efficient deployment of its workforce. A first analysis of the mid-sized spare parts section of the showed that the picking performance per person had increased by some 10 percent.

UNDISCOVERED POSSIBILITIES

Besides the already mentioned examples, there are a lot of further possibilities for improving processes in distribution centers and, often, the responsible persons are not aware of the improvement potentials

For companies, there are many starting points, from where their intralogistics can be optimized via easy measures and resources. Appropriate software solutions can allow every distribution center to individually analyze, calculate and activate hidden potentials.