Simplifying A Complex Network Model

The warehouse is usually the last place where a product is kept before leaving for its journey to the final customer. Therefore, it is a synonym of money.

Only a very agile operation, which has the right products at the right time, can cater to the increasingly shorter and demanding customer delivery times. Reception, picking and put away operations must be perfectly aligned and coordinated to avoid extra work hours, over-occupation of the means of material movement, such as pallet stackers or forklifts, and over-searching due to information errors about the real product locations.

These operations do not survive by themselves - they can only do so with good stock management in place. We might have a great picking process, but it will only be successful if the right SKUs and quantities are available. Plus, it avoids the stagnation of cash flow due to obsolete stock with months or years of coverage that can lead to the unnecessary urgency of acquiring more warehouse space.

The company

This company is a leader in paper and office article distribution. In 2019, the company completed the fusion of two companies in the German market to extend its operation and market share.

This acquisition has led to the need of aligning all the information and standardising the whole process so the operation can continue to run smoothly after the transition period.

However, before the fusion, transport costs that used to represent 9.3% of the total invoice have increased to 13% in 2021.A similar trend happened with warehousing costs – from 12% to 21%.

To maintain a profitable business, the company needs to understand where this cost is and how it can be reduced by increasing the synergies of the network.

The challenge

Currently, the company has 17 warehouses through the country. There are three main hubs: two in the north and one in the south. Each warehouse must fulfill the demand of its pre-defined region. Despite the decentralisation of stock, there is an intra-logistic transport network which is used as a response to:

• Choice of stock allocation made by the commercial team.

• Issues with stock availability at preferred warehouses.

The average fleet occupation stands at around 70% across different sites, the consequence of almost fully interconnected network in place to respond to virtually any demand scenario and stock transfer requirements.

For local demand, each warehouse has a fleet of small trucks that perform pick-ups and deliveries directly to the customer .Customers are getting about 60% service levels because of the compound effect of 90% service level from intra-transport and then last-mile deliveries.

The Approach

1. Redefine the keeping and buying stock strategy

The ABC XYZ analysis is fundamental to optimise inventory. It starts by classifying each SKU based on its sales volume(ABC) and frequency (XYZ), which is a proxy to risk due to fluctuations.

This approach separates the SKUs into nine different classes that enables us to create a strategic rule to each category. For example, BZ, CY, CZ references should have very low stock levels and the strategy should be buy-to-order to avoid becoming obsolete thus increasing turnover.

2. Review stock replenishment algorithm

The stock replenishment algorithm must be fine-tuned so that it correctly balances service levels and stock coverage. In this exercise, we considered the global demand for the whole network and built two different scenarios: one that uses the data for future demand and the other that looks to past demand. The calculations were made for a period of three months, day by day, which accounts for any kind of demand variability that may appear.

The first one leads to almost zero stockouts, but it relies on a very good sales forecast. The second one is moreconservative, but still shows a great balance between a decrease in stock levels and an improvement on service level.

The stock evolution and comparison to current stock levels of a single reference can be seen below:

3. Improve fleet occupation

The trucks used for last mile deliveries have very low occupation, both in time and weight. In this analysis, it is important to separate the driver and truck opening time. A driver might be available during his shift (8h), but the truck is available for the whole day,or at most, during the time window for delivery set by the clients.

The average occupation in terms of time was33% which illustrates this effect. In the graph below, we can see that mostroutes last for a period of 8h (red line). The opportunity here is to create two shifts to maximise the customer time window (blue line) and consequently the vehicle time occupation and number of deliveries performed per day.

4. Design the best transport model

To study which transportation model should be adopted, we started by classifying the main suppliers and by quantifying the demand volume for each region. Different scenarios were set to understand where each supplier should deliver and, by consequence, where the stock should be placed. Each scenario considered the total number of kms travelled and the number of trucks needed as well as a penalty for every trip performed with sub-occupation.