Bunnik Plants B.V. is a Dutch plant grower that struggled to deliver their products on time to the customer, because of changing customer desired. By using the Enterprise Dynamics software package a simulation model was constructed that aided in the decision making process on how to redesign the daily order delivery strategy. With the model both duration and capacity analyses in relation to transportation were performed that resulted in valuable insights on to what extent the current order delivery process could be altered.
Bunnik Plants B.V. is an internationally oriented plant-grower, located in Bleiswijk, the Netherlands. The company has a total space of 25 hectares, divided over 7 locations, on which approximately 100 different plant species are grown. On average 600.000 plants are being produced every week. Furthermore plants are imported from Asia, Central America and the Mediterranean. In consultation with the customer customizable product assortments can be realized, by adding customer specific attributes to the ordered plants such as pots, labels, stickers and seal packaging material.
The company is part the Dutch floricultural value chain, which is the main supplier of both potted plants and flowers in Europe. The chain is characterized by its fast product throughput, which is necessary given the fast quality loss due to perishability of the products distributed. Within the chain, plant-growers sell their products either directly to wholesalers, or use the auction as an intermediary sales platform. Although it is preferred to sell to the customers directly since there is a higher profit margin to be gained, products are also sold via the auction to retain customer awareness, and to get rid of excess capacity that cannot be sold via regular sales.
Project objective
Bunnik Plants recognized the need to create a faster and more efficient order delivery process, to cope with the current developments within the sector, and to protect their market share from competitors. In order to decide on how to reorganize their regime, the grower needed more knowledge on how delivery related processes affected the companies' performance, and what would be feasible reconfiguration options to consider, given specific KPIs. A simulation model was set up using the Enterprise Dynamics software package, to determine what optimal truck departure deadlines should be set to meet the required service level, and whether it was possible to transport all products with a new suggested delivery configuration, given the available truck fleet.
Model functioning
The model mimics the order handling process of Bunnik Plants. Orders that arrive throughout the day are modelled in order batches, which are being released at three specific moments per day. Once a batch is released, customer location tags are added to each individual order for the transportation procedure later in the model. Order production is then initiated. After production the orders are ranked in a queue based on their locational tag, and a predetermined priority list on which orders to deliver first. The products are then loaded in the trucks using a so called first-fit bin packing algorithm: Orders can only be loaded in a truck, when there is enough space available. If an order could not be loaded in a truck anymore, a new truck is assigned to load the current order. After loading of an order, the next order in line is loaded until all orders have been assigned to the trucks used for transportation. With this step the amount of trucks required is determined, and route sequences are set based on the products that are loaded in each truck. After loading, the trucks departed for delivery.
For the delivery of the products, two networks are set up, each network representing a clustered region of customer delivery locations. A network contains two types of delivery locations: Distribution centers, where most of the customer orders are being delivered, and specific locations of customers that have set special delivery arrangements with Bunnik Plants, or cannot be supplied using one of the distribution centers.
Data to drive the model is retrieved using two main data sources. All information on order arrival, order size and customer location is gained from an internal order handling database. With this information specific order patterns and quantities can be found, as well as a seasonal ordering trend. Information about the driving times between locations, and unload times at these locations is deduced from trucks' GPS-tracking information. Both the individual input datasets, as well as the model output data were validated before output analyses were performed: Test runs were performed for which the output was collected. This output was compared to the model input, which was based on historical data retrieved from Bunnik Plants' IT-systems. Once a close match between both data sets was obtained, the model was perceived valid.
Analysis
To determine the optimal truck departure deadlines, first all simulated truck delivery times were sorted based on the amount of delivery locations that had been visited within a region. For each delivery run, with respect to the number of delivery locations, the maximum delivery duration time was measured by determining for which duration time 98% of these trucks runs simulated delivered the products to the customers. Subtracting these times from predetermined fixed delivery deadlines showed what the optimal truck departure times should be
During the time this project was performed, a new delivery strategy setup was proposed, for which the changes on truck fleet usage needed to be analyzed. An adjusted model was constructed to meet the new strategy specifications, and the model outcomes on truck fleet usage were compared to the original model outcomes. With the new model it was predicted what the effect on the daily allocation of the truck fleet would be when less daily delivery moments were used, and whether this truck fleet was sufficiently large to be able to deliver the products to the customers throughout the year.
Results
With the performed analysis on delivery duration time, Bunnik Plants gained insight in the amount of time that is needed to deliver their product to the customer, given a desired service level. Both a minimum delivery time for each cluster of customers was obtained, as well as the average time required given the number of deliveries that had to be performed by a single truck.
The analysis on truck fleet usage revealed that Bunnik Plants was capable to deliver almost all of their products to the customers throughout the year, when less daily delivery moments were being used. Using less delivery moments resulted in more products that had to be transported at a specific time of the day, but it was predicted that this would not lead to overcapacity problems in the near future.
Both outcomes aided Bunnik Plants in the development of a new order delivery strategy that better suits the changing desires of its customers. Simulation studies proved its usefulness since the outcomes on important KPIs could be predicted, without having to implement a new scenario first.
Rene Vandewall, MSc in Management, Economics and Consumer studies - Wageningen University.
