Allocation Comparison AS-IS vs OPT
6. Unit‐load Warehousing
6.1.2 Operations scheduling
The first step deals with the determination of the optimal lane depth per each SKU considering a specific period. The principle issues related to the application of the illustrated pattern consists on the fact that is bases on a constant demand rate. Therefore, the pallets of a generic SKU filling a lane are assumed to be shipped with a constant predictable rate. In real warehouse, the inventory differently turns for different SKUs and the honey combing and accessibility costs effectively depends on the inbound as well as outbound flows. As instance, the honey combing cost for slow moving SKUs, which hold the lane for longer, is higher than for fast‐moving SKUs.
The second relevant issue consists on the fact that the adoption of the pattern of a period suggests the proper number of lane of the optimal depth to devote to each SKU, but consider all SKUs as stored at the same time. Finally, the allocation of lanes suggested by the pattern gives an instant picture of the best fitting layout, without taking into account the operative inbound and outbound flows. Attempting to tackle these critical aspects, the procedure continues with the second step named dynamic scheduling. This step enables the dynamic scheduling of inbound and outbound operations by considering an interval of time for which a set of lanes is entirely devoted to a SKU. This time, also named release interval, is the period from the end of the put‐away process (i.e. due to manufacturing or inbound receiving) until the retrieval of the last pallet stored in the lane. This is the interval allowing to assign a set of lanes, of the proper depth, to a batch of a generic SKU, and represents also the interval for which a lane is held and after then released. Therefore, the overall layout resulting by the sum of the lanes, of the proper depth, devoted to each SKU, for the whole population of SKUs, depends on the considered release interval. Furthermore, assuming such time batch as the average turn rate of each SKU, the decision‐maker can address seasonality, demand rate of different SKUs, and obtain a more accurate metric of the whole storage capacity of the system.
Figure 69. Layout configuration
This approach allows extending the static pattern proposed by Bartholdi and Hackman (2011), by considering also the interval of time after then a lane is emptied and released. By computing, day by day, the overall storage capacity (i.e. in terms of both locations and lanes) required to store the incoming lots of SKUs, the procedure provides also a useful tool to study and design the storage layout. Indeed, the warehousing system, treated as an incapacitated queue system, experiences day by day a ranging holding capacity, rolling accounted in accordance with the incoming lots of SKUs and the related lanes releases after the release interval.
Figure 69 gives a picture of the storage layout resulting by the adoption of the first two steps of the top‐down procedure. The incoming of production lots (i.e. batches) of SKUs allow the setting of the optimal lane depth per each SKU, whilst the adoption of the dynamic scheduling module provides the ranging holding capacity of the storage system to fit day by day put‐away and retrieval processes.
6.1.3 Simulation
The third step deals with the simulation and related assessment of the space efficiency performance of the unit‐load warehousing system. Such performances regard with the ratio of occupied and unoccupied storage locations to the total capacity, the ratio of open (i.e. occupied) lanes to the overall number of lanes, and the saturation of every open lane. In order to realize a benchmark of the storage system, it is necessary to establish and set a storage layout, composed by the sum of the number of lanes of each depth, and then simulate how the system matches the historical inventory or the historical inbound and outbound flows. Once the layout is defined (through the first two phases of the procedure), the measurement of space saturation performances results by the process of filling the available lanes and storage locations with the historical inventory or inbound and outbound flows.The assignment of incoming lots of SKUs to the lanes set through the first two steps is based on a greedy heuristics, consisting on both rankings of incoming lots of SKUs and available lanes of a generic depth. The first arriving lot of a generic SKU occupies the required lanes of the optimal depth, if available, decreasing the overall number of such lanes, until the last pallet of that lot is retrieved. If the saturation metrics and the space efficiency performances do not satisfy the decision‐maker (i.e. low values of space saturation), the procedures allows to iterate the step of dynamic scheduling in order to set other release interval and to rearrange the layout accordingly.
Figure 70. Design top‐down procedure
Figure 70 represents the flow‐chart of the main steps of the proposed procedure. The preliminary phase regards with enterprise data collection and is based on the gathering of the information on warehouse inventory and throughput to be store in properly defined database architecture. The first step regards with the settings of the horizon of analysis, the storage mode to study (i.e. floor storage, drive‐in, drive‐through, flow‐rack, ASRVS) and the layout and infrastructure as input for the application of the lane depth patterns. The pattern adoption defines the optimal lane depth to allocate a set of SKUs, considering an interval of time. The pattern, if properly arranged, responds to Start Storage mode Setting Analysis horizon Setting Layout & infrastructure Setting Optimal lane depth patterns Operations Scheduling Static Layout Dynamic Layout OR Lane depth Setting Layout scenario Setting Layout Scenario Simulation End Feedback 1 Feedback 2 Feedback 3
multiple storage modes implemented in the procedure (i.e. floor storage, drive‐in racks, drive through racks, flow‐racks, ASRV systems).
The second step enables to extend to static pattern considering the delay between put‐away and retrieval processes. This step considers the time for which a set of lanes is on average held by a SKU, and through a rolling approach, gives a picture of the overall storage capacity of system (i.e. in terms of locations and lanes) required to match inbound and outbound flows. Two main functionalities are available. The so‐called static layout inherits the optimal lane depth computed for the average inbound lot of each SKU, and holds the proper number of lanes for the average storage time of a SKU. Conversely, the so‐called dynamic layout computes different optimal depths for each different inbound lot of a generic SKU and then assigns the proper number of lanes for the average storage time of a SKU. The output of both functionalities consists on a set of layout configurations, rolling day by day, characterized by different value of storage capacity.
The third step sets the specific layout configuration, among the range of the configuration proposed by the dynamic scheduling module, and enables simulating and assessing the space efficiency performances over real inbound and outbound historical flows.