The issue of material handling involves the design and operative control of warehousing systems (i.e.

, distribution centres), which allow

matching vendors and demands, smoothing with seasonality, consolidating products and arranging distribution activities. Warehousing
systems play a crucial role in providing efficiency and customer satisfaction. The warehouse design entails a wide set of decisions, which
involve layout constraints and operative issues that seriously affect the performances and the overall logistics costs.

This study presents an original decision-support system (DSS) for the design, management, and control of warehousing systems.
Specifically, the proposed DSS implements a top-down methodology that considers both strategic warehouse design and operative operations
management. The DSS can simulate the logistics and material handling performances of a warehousing system. Heuristic methods and
algorithms address several critical warehouse issues, such as the order picking process, which is responsible for 55% of the overall costs in a
distribution centre. The benefits due to the adoption of the proposed decision-support system are summarised as a dashboard of key
performance indicators (KPIs) of space and time efficiency that allow logistics providers, practitioners, and managers as well as
academicians and educators to face real-world warehousing instances and to find useful guidelines for material handling.

1. Introduction and background

In recent years, enterprises have completely reconfigured their supply chain to address increasing customer service
levels and demand variability. Warehouses play a pivotal role in the supply chain, and requirements for warehousing
operations have signifi-cantly increased. Specifically, the customer needs in terms of the order accuracy and response time,
order frequency, order quantity and order size have dramatically changed with the global economy and new demand trends
(e.g., e-commerce). The literature has widely debated the issues of warehouse design and management, which is aimed at
minimising the operation costs and time and increasing the supply chain performance. Comprehensive surveys on
warehouse and industrial storage system topics have been proposed by De Koster et al. [1], Gu et al. [2] and Dallari et al.
[3].

The main function of the warehousing systems is to receive products (from inbound or manufacturing lines), to store
materials until they are requested, and then, to extract products from inventory and ship them in response to the customers’
orders.
Fig. 1 illustrates a conceptual framework for classifying warehouse operations, considering the definitions of entities,
processes, activities, and decisions as related to storage systems.
Products typically arrive in large units, such as unit-loads, and standard or custom containers, or pallets, which cause
the related labour and handling activities to be less expensive. Incoming products must be put away, which is the most
significant warehouse function. The put-away process entails a set of interdependent decisions [2]: given a warehouse
configuration (based on the layout parameters of Fig. 1), how much inventory should be held for a generic SKU (the so-
called allocation in Fig. 1), and where should it be stored (the so-called assignment in Fig. 1)?

The warehousing system pursues the transformation of the large and relatively homogeneous arrival materials into
small, frequent and heterogeneous output quantities in response to customer demands. The small and frequent output
quantities result from the fulfilment of the customer order lists.
Order picking is one of the prime components of labour and warehouse-associated costs. Two alternative configurations
of layout types are common for picking. One, the so-called multi-level picking (see Fig. 1), executes high-level picking
directly from storage locations, which are all accessible by picking equipment (e.g., turret-trucks). The other, the so-called
forward-reserve

(see Fig. 1), executes low-level picking from the easily accessible forward area, which holds the bulk of the inventory for
every product in a larger, reserve storage area. When given a product for which the inventory is low in the forward area,
replenishment is realised from the reserve. For an exhaustive description of the patterns that are depicted in the proposed
framework, a definition of zoning, batching and routing in warehousing is given. The zoning comprises partitioning the
warehouse into different zones, which correspond to work stations. Pickers are assigned to zones, and workers
progressively assemble each order, passing it along from zone to zone. The batching comprises making a picker retrieve
multiple orders in one trip. Even though batching represents a very useful approach to reduce travelling, it requires the
retrieved SKUs be sorted into a single order. Lastly, the routing defines an appropriate sequence of items on the order list to
ensure a good route through the warehouse.Overall, two main aspects lead to enhanced performance: the warehouse design
(1) and the operations control (2).
The first aspect refers to the layout constraints and parameters (illustrated in Fig. 1), the storage equipment and the
high-level strategic decisions on the total inventory of the facility. The second addresses the warehouse operative activities,
such as put-away, replenishment and order picking, focusing on models, techniques, and methodologies to enhance the
operative performances (e.g., zoning, batching, routing). These two aspects significantly affect warehouse performances
and have a direct influence on the level of service of the overall logistic chain (i.e., the steps before and after the
warehousing system of Fig. 1).

The literature proposes a wide set of warehouse KPIs that include the throughput capacity (the material flow processed
through the warehouse per time unit), the storage capacity, the response time (the time within the order arrival and its
shipment), the cost rate, and the cost per unit of material flow shipped by the warehouse. All of these metrics are affected
by the management of space and time, which are critical for every logistic process.
 Generally, the contributions of the literature address the problem of warehouse design rather than the management of
warehouse operations separately. Gu et al. [2] describe inbound/ outbound processes and review the literature, classifying
the papers on the basis of the scope of analysis, the adopted method and the type of the observed warehouse (e.g.,
automated, conventional multi-aisle storage systems).

Typically, warehousing problems are non-polynomial (NP) problems and have a very large amount of real-world data
to manage. Therefore, user-friendly and timeless solutions for the warehousing issues are ambitious aims for computer-
based applications.

The remainder of this study describes the conceptual design and development of a decision-support system (DSS) for
the strategic design and the management of operative activities in a warehousing system. Specifically, it supports the design
of complex multi-zone forward-reserve picker-to-part storage systems and provides multi-scenario simulation for KPI
assess-ments. The DSS implements sets of heuristic methodologies to support data-oriented analyses and performance
enhancement.

The management and control of warehousing system (i.e., industrial storage system) activities and processes range
among various design alternatives and involve different expertise. For example, the problem of layout design, the definition
of the total storage capacity, the determination of the number of aisles, the types of racks, the locations of the products (i.e.,
stock-keeping-units or SKUs) within the storage area, the stock per each SKU, and so on, involve interrelated areas and are
challenging but can be addressed through a unique modelling formulation. The majority of the contributions reviewed in
the literature [1–3] focus on a single aspect of the warehousing problem, thereby ignoring the integration of multi-purpose
approaches.

The proposed DSS develops a top-down methodology for the comprehensive design of a warehousing system that
allows for the decision-maker to develop and compare different configurations and scenarios in a user-friendly computer
environment. It implements multi-scenario simulation techni-ques to address real-world case studies, to highlight the
interdependency among decisions and to identify useful guide-lines about warehousing issues.

DSSs are computer-based tools that have been adapted to support and aid complex decision-making and problem
solving [4,5]. Research in this area typically highlights the importance of information technology in improving efficiency
adopted by users to make decisions, improving their effectiveness [6,7]. Specifically, the literature reveals the benefits of
using computer-based systems to support logistics management, especially in the areas of logistics, transportation, and
warehousing [8–10].

Rouwenhorst et al. [11] and Svestka [12] develop interactive decision support systems that are aimed at the conceptual
design of dedicated storage systems to store and retrieve pallet loads (i.e., unit loads). Other studies present tools for
managing order picking systems (OPS) (i.e., less-than-unit loads), which support the analysis of operating data (e.g., the
SKU master file, order master file, inventory master file) to determine the requirements for the OP operations and storage
capacity [13,14].

Currently, the literature does not provide any contributions that can combine warehouse design and operations patterns
into a unique analysis, as suggested in the proposed DSS.
The proposed DSS is written in a high-level programming language (C#) that utilises a relational database that can
gather, store and manage datasets from a real-world warehousing instance. Warehousing systems generally collect tens or
hundreds of thousands of SKUs, with customer demands of millions of order lines per year, while managing inbound-
outbound processes, quality checking, and scheduling shipments. For this purpose, industry invests in the development of
integrated information solutions, which are referred to as warehouse management systems (WMS). These commercial
systems provide a real-time view of material handling, often advising the efficient use of space, labour, and equipment
[15]. Nevertheless, WMS solutions com-prise management systems that have no functionalities that are related to decision-
making on warehouse design and optimisation.

The lack of systemic methodology on this topic highlights the need to provide a DSS that can gather data from real-
world instances and implement sets of effective heuristics to rapidly support decision processes on warehousing design and
manage-ment. The aim of this study is to illustrate an innovative architecture of DSS for the analysis of warehousing
systems while considering the layout features, storage equipment, allocation and assignment problems, adopting numerical
simulations to assess results, statistics and performances.

The expected results of the proposed computer aided system can be exploited by disseminating knowledge among
logistic providers, practitioners, and managers, by educating and improv-ing industrial engineer expertise and by analysing
real-world case studies.
 The remainder of this study is organised as follows. Section 2 presents a design-support methodology for warehousing
sys-tems and reports a more relevant definition of the main leverage of analysis. Section 3 illustrates the developed DSS
functionali-ties through graphic user interfaces (GUIs) and the data-management section. Section 4 gives a picture of the
potential results and analyses that were conducted through the applica-tion of proposed DSSs to real-world industry
instances. Lastly, Section 5 discusses the conclusions and provides directions for further research.

2. Solving warehouse design and management issues: a top-down procedure

The proposed DSS implements a top-down procedure for the design and management of a forward-reserve OPS, as
illustrated by Accorsi et al. [16]. This methodology organises procedures, models, and algorithms in an organic sequential
decision to provide a wide set of solutions for storage layout, storage allocation, and storage assignment. The decision-
maker conducts a sequence of analyses, generating sets of alternative warehouse configurations to be assessed through a
what-if multi-scenario simulation. The goal performance is the minimisation of the total travelled distance due to picking,
which represents 55% of the overall warehouse costs. The reduction in the distance means a reduction in the number of
travelling material handling solutions (e.g., forklifts and guided vehicles) that are necessary to move materials, a reduction
in the vehicle congestions, parking areas, costs of travelling, mainte-nance, labour, and other outcomes.
Thus, the feedback flow illustrated in Fig. 2 allows for the user to rearrange his/her decisions to achieve efficiency in
both the warehouse design and the operations. The following sub-sections focus on the main decision steps.

2.1. Layout

The first decision involves the design of the warehouse layout. The study of a warehouse layout is based on the
assessment of the facility storage capacity. The proposed methodology is based on the historical inventory and customer
demand (or the demand forecasts) that are assumed as input for a stock out risk evaluation analysis that is aimed to
establish the required storage capacity of the warehousing system (i.e., designed from a green-field).
The purpose of this step is to set the facility layout through the definition of a set of parameters (see Fig. 1), such as the
shape factor, the number of aisles, the number of bays per aisle, the rack sizes and types, and the characteristics of the unit
load (i.e., the pallet size or other container solutions). The DSS rationalises the whole storage space by devoting different
zones for different SKUs in terms of the shape and size, which likely require specific racks or equipment.

2.2. Allocation

The storage allocation strategies establish a fraction of the overall available storage space in the forward area to be
devoted to the generic SKU according to specific criteria, given a certain time horizon. An equal space (EQS) strategy
devotes the same fraction of space to each SKU, while an equal time strategy (EQT) ensures the same number of restocks
for each SKU given a selected time horizon. Both of these strategies are renowned in industry and are suitable for every
storage context (i.e., cartons-cases picking). The optimal strategy (OPT), as proposed by Bartholdi and Hackman
[17] and previously by Hackman and Rosenblatt [18], minimises the restocking to the forward area for pieces-picking
(i.e., the order picking for the small parts). The proposed DSS implements different so-called allocation strategies by which
to configure alternative scenarios of stock for every SKU. In the forward area (i.e., the fast-pick area or the low-level
locations), the choice of the stock level to devote to each SKU affects the replenishment activities as well as the picking
processes because this choice influences the locations of the SKUs [16]. The system also supports a pattern [17] to
determine the sub-set of SKUs that maximise the net-benefit of the forward area, considering both the time savings per pick
(i.e., the pick from the forward vs. the pick from the reserve) and the time for replenishment.

At this step, the decision-maker matches the allocation results with layout features and eventually considers the
opportunity to return to the top for re-layout planning.

2.3. Assignment

The storage assignment strategies establish the appropriate locations to assign to the SKUs in accordance with different
heuristics. The DSS asks the behaviour of selected SKUs within the demand profile for a selected time horizon.
Information on the picking processes is collected to compute a panel of metrics used for SKU classification. Specifically,
an index-based assignment policy classifies the overall set of SKUs according to the effective criteria as the popularity (P)
(i.e., number of requests per each SKU), the turn-over (T) (i.e., the ratio of the demand to the inventory for each SKU), the
cube-per-order index (COI), the order closing (OC) (i.e., the ability of an item to close an order). For a detailed description
of such heuristics and methods for index-based assignment policies, a recent literature contribution [16] is recommended.

Another relevant aspect that can be considered through the proposed DSS is the correlation among the SKUs that are
requested together by customers. Correlated-based assignment policies can be applied to group SKUs that are requested
together and assign them to storage locations that are close to each other, to save on the travelling needed for the picking
activities. The implemented correlated-based approach comprises the following three main steps:

Correlation analysis. The level of correlation is generally measured by introducing a similarity index among the SKUs.
This procedure allows comparing general-purpose similarity indices, e.g., the Jaccard index proposed by McAuley [19] and
certain problem-oriented issues.

Clustering. This step concerns the adoption of hierarchical clustering algorithms (e.g., single linkage, complete linkage,
group average) and different similarity-cut thresholds of a dendrogram (i.e., value-based, percentile-based) [20–22].
Cluster Assignment. This step computes the above-mentioned metrics (i.e., popularity, turn-over, order closing) for
each cluster of SKUs (e.g., the popularity of a cluster is given by the weighted sum of the popularity of the included SKUs)
and to sort the clusters of SKUs accordingly, as summarised in Accorsi et al. [16].
Regardless of the adoption of specific assignment policies (i.e., index-based or correlated-based), the assignment step
returns a list of SKUs (or a cluster of SKUs) that are sorted in accordance with the selected criteria, to be properly matched
with a list of locations, ranked by the increasing value of the single-command (SC) path (i.e., the distance to visit a location
from/to a shipping/receiving dock). The computation of the SC depends on the location of the shipping and receiving docks
(e.g., left/right corner, distributed, same side, different side) and on the so-called aisle-visiting strategies (i.e., mono- and bi-
directional aisles). To fit the layout constraints multiple (i.e., approximately twenty), configurations of both aspects are
implemented. Lastly, each SKU is assigned to the most convenient available location in accordance with a greedy heuristic
approach.

2.4. Multi-scenario simulation

The set of decisions that were previously addressed by the DSS (i.e., the layout design, allocation and assignment)
provides a specific configuration for a warehouse scenario. Multiple iterations of the DSS allow for generating multiple
warehouse scenarios, which differ in their layout configuration, storage allocation, and/ or storage assignment criteria. In
conclusion, a what-if multi-scenario simulation of operative performances (i.e., travelling for put-away, replenishment and
picking) enables the decision-maker to assess the best solution for the warehouse design and management by the
minimisation of the total travelling distance, time and cost.

3. DSS functionality and design

The DSS provides a useful and user-friendly tool for managers and decision-makers who have no background and
expertise in programming and software development but who frequently face warehousing system design and operations
issues. The DSS implements database management system (DBMS) architectures for data storage, models and heuristic
algorithms and user-friendly graphical user interfaces (GUI) that enable interactive queries, reporting and graphic
visualisation.

The proposed application is based on a stand-alone database. Decision-process inputs with regard to operative features,
costs, and other parameters are generally handled by practitioners in warehouse operations, whilst outputs comprise
operative KPIs that are usually tracked in the real world (e.g., the pick-rate, time/ travelling for picking). The SQL database
architecture enables users to gather, store and manage a very large amount of data quickly, which is gatherable by users
through dynamic queries. Further-more, graphical 3D views of warehousing scenarios are drawn automatically by an ad-
hoc graphical user interface with Auto-CAD1.

The application is organised around a main GUI that presents all of the principle features and commands to load data or
projects and to save results. The tool enables the following main functionalities:
Design a new warehousing system (we call green-field).
Import the existing layout (we call brown-field) to perform an allocation-assignment analysis.
Run the DSS for a complete layout-allocation-assignment analysis of a generic warehouse zone (i.e., in accordance with
warehouse zoning).

Merge single-multiple warehouse zones (i.e., in accordance with warehouse zoning) as an aggregated system.
Implement heuristics for storage allocation, assignment, single-order picker-routing, order-batching.
Develop a what-if multi-scenario analysis for the warehousing KPIs.

Draw a graphical 2D/3D warehouse in agreement with different designed scenarios.

3.1. Database considerations

 The DSS utilises the aggregated historical data that is stored in the database as the foundation for the application of all
of the heuristics, methods, and analyses. This section focuses on the information and data architecture as a basis of the
proposed DSS. In warehouse operations, the very large amount of data that is to be handled is critical. Warehousing
systems manage tens of thousands of SKUs that are picked from thousands of locations to fulfil thousands of demand lines
per day (see, for example, spare parts storage and management systems in the automotive industry). Warehousing activities
are usually tracked by enterprise WMSs. The preliminary step of analysis comprises filtering the available historical
information (e.g., the SKU master file, inventory, demand) to build a comprehensive stand-alone data-base in accordance
with the entity-relationship (E-R) diagram illustrated in Fig. 3.

The developed DBMS represents an interface between the data and the decision-maker. This system involves the
processing of a considerable amount of data (see Fig. 4), which is necessary to describe univocally the characteristics of the
warehousing system.

This system comprises a relational SQL architecture that is powered by Access TM but is quickly replaceable by any
other commercial DBMS (e.g., MySQL1, DB21). The database includes a set of tables (see Table 1) that allows for a
comprehensive description of the system’s object of analysis through a typical snowflake structure. Meaningful preliminary
studies on the unified modelling language (UML) and E-R diagrams are crucial to designing an informative architecture
with the tool and to aid in further code maintenance of the modifications [23,24].

This database architecture has various advantages. First, it enables users to track the inventory and to localise a generic
SKU in both the forward and reserve storage areas. On the other hand, it allows for a wide set of dynamic views and
queries to create a performance dashboard of the warehousing system.
The client side comprises a user-friendly interface made by GUIs. The decision-maker plays opportunities to design a
warehouse zone from ground-zero (i.e., a green-field scenario), to add a new zone to an existing warehouse (called here an
‘‘expansion scenario’’) or to import an existing storage zone for an allocation-assignment analysis (i.e., a brown-field
scenario).

A what-if multi-scenario simulation of put-away, replenish-ment and order-picking and outbound (e.g., picking)
activities is performed as a benchmark to assess the efficacy of each scenario and the effectiveness of the adopted
allocation and assignment policies. To enhance the picking performances, an order-batching algorithm (i.e., whose
description is not in the scope of this paper) based on a clustering approach is implemented as a batching tool, and a
travelling salesman problem (TSP) heuristic (i.e., nearest neighbour) is developed as a routing tool.

3.2. Graphical user interfaces (GUIs)

GUIs enable the user to conduct analysis and to lead decisions through the DSS. The main window presents a toolbar to
load or save a project. During any execution run, the user specifies the domain and dataset object of analysis. Statistics and
results are summarised on the bottom of the control panel as a quick report window to inform the decision-maker about the
computer processing. For each project, multiple ware-housing scenarios can be developed. A what-if experimental analysis
based on a dynamic simulation can be conducted to compare the performance of the warehousing system under different
configurations and operating conditions. At the end of each simulation, the obtained results and KPIs are depicted and
stored into the appropriate tables of the database (see Table 1). One of the advantages of saving the results of each run is
the opportunity to draw out effective guidelines for the design and management of complex warehousing systems. The GUI
comprises distinct modules that are further detailed in the following sub-sections.

3.2.1. Layout GUI

To begin the design of the warehousing system, the decision-maker sets the total warehousing holding capacity. Given a
dataset on the historical demand or inventory, the user must guarantee the overall level of the stock and properly organise
the available space. The leverage handled through the proposed DSS for the configuration of the layout are the shape
factor, the unit load sizes, the unit load location, the base module sizes, the aisle width and number, and the rack types, as
illustrated in Fig. 5

This GUI reports the characteristics of the layout configuration, such as the total storage capacity, the numbers and
sizes of the aisles and bays, the storage saturation (i.e., the ratio of the storage volume to the overall available volume), the
number of SKUs stored per each aisle or per each bay, etc.

Once the warehouse is designed (or imported), the DSS computes the three-dimensional coordinates of all of the
locations and stores them into the database (see Fig. 3) for further simulation analysis. The DSS even includes an
AutoCAD1 application, which adopts real-world rack libraries to make a bi-dimensional and three-dimensional picture of
the warehousing system.
Specifically, this application allows for analysts and logistics providers to import rack components (e.g., beams,
columns) that are taken from manufactures’ catalogues and configure a truthful and accurate warehouse layout. The system
evaluates the maximum load weight of the inventory and checks for the availability of appropriate rack components that are
suitable in their sizes and characteristics. Fig. 6a gives a picture of some three-dimensional views of warehouses that result
from the DSS application. As a result, the detailed list of parts is given as a rough estimation of the total investment.

3.2.2. Allocation GUI

This GUI allows for the user to compare different allocation strategies that were attempted to allocate the appropriate
storage volume to a generic SKU within the forward area for a typical forward-reserve picker-to-part OPS. The DSS
encompasses four main allocation strategies, three of which were previously described in Section 2.2; one is hereby
proposed, the so-called EQT*, which allocates the appropriate storage space to each SKU considering the demand in terms
of the retrieved volume and pick lines. This module has an open architecture that eventually allows an easy implementation
of other additional allocation strategies.
Fig. 7 illustrates the GUI such as is proposed to the decision-maker. On the left, two input command windows (‘‘Rack
Level (n.)’’ and ‘‘Allocation Strategy’’) are presented to define the number of rack levels that are devoted to the forward
area and to select the allocation strategies by which to adopt. Thus, the user has the opportunity to configure a low-level or
high-level picking system and assign the highest levels to the reserve storage area.

The calendar panel (on the left of Fig. 7) selects the horizon of analysis, by filtering the dataset through dynamic SQL
queries. Different time batches are selected to compute the fraction of the storage volume that is devoted to each SKU
according to the historical demand and inventory data. For example, given a temporal batch (from August 31st, 2011 to
September 28th, 2011), a panel of allocation strategies accordingly allocates to every SKU the storage volume, cartons, and
unit loads within the forward area (see the tables in the middle).

Storage space is often a precious resource to be handled, to reach efficiency and reduce operating costs. At this step, the
decision-maker can evaluate the net benefit of the forward area, according to the pattern that was briefly introduced in
Section 2.2. The sub-set of SKUs, which maximises the net-benefit of the forward area, corresponds to the maximum value
of the curve depicted in Fig. 7.

If an existing warehouse zone/system is imported and loaded, the AS-IS inventory per each SKU (i.e., the number of
cartons and unit loads in both the forward and reserve areas) is known, and related data are stored into the database. Thus,
the user can skip the allocation module, which is not considered as leverage of the analysis, leaping from the layout design
module directly to the storage assignment problem.

3.2.3. Assignment GUI

This GUI leads the decision-maker towards the assignment issue by the definition of the appropriate location to assign
to a generic SKU in the forward area. Considering the horizon of analysis (i.e., the same chosen for allocation analysis or
different), the user classifies SKUs according to a set of proposed criteria or metrics (i.e., the index-based functionality), to
assess the correla-tion among the SKUs (i.e., correlation-based functionality) through a clustering approach.

Both of the opportunities compute a ranked list of the SKUs (eventually computing clusters of SKUs), respond to
specific criteria (see the previously cited popularity, turn, and order closing), to be properly matched with a list of locations,
according to the procedure presented in Section 2.3. More than twenty combinations for the sites of the shipping and
receiving docks (e.g., corner, middle, bottom-up) that affect the single-command time to access a generic location are
considered.

Once the appropriate location in the forward area is assigned to a specific SKU, the reserve area is accordingly arranged
by the adoption of greedy heuristics to reduce the distance between an item and its reserve.

The results of the assignment module are store into the database and are roughly illustrated as the bird’s eye view of the
designed warehouse zone. The bird’s eye view is a frame shot of the SKU locations, where each SKU is differently
coloured, and the storage details (e.g., the location code, item code, and number of cartons per item) are summarised. The
DSS also fills the rack with the SKUs in the designed layout in a three-dimensional view (see a sample in Fig. 6b). By
considering real commercial racks, the decision-maker obtains a ready-to-print version of the designed warehouse that is
useful for equipment and systems manufac-turers and providers as well as warehouse operators who are responsible for put
away and picking activities.
3.2.4. Simulation GUI
In warehousing operations, different categories of SKUs in terms of the shape, volume, weight or size of packaging are
assigned to different zones, adopting different types of rack or storage equipment according to a zoning approach. Several
configurations for the storage zones, separately and independently designed through previous GUIs, are hereby saved and
are further selected by the decision-maker to be merged into a unique system (see Fig. 8a).

This GUI enables us to configure articulated and complex warehouses that are made by different storage zones, as
commonly occurs in real-instance warehousing problems (exem-plified in the literature by [25]). Furthermore, this GUI
matches the decisional steps with what-if simulation analysis. By setting the layout (i.e., merging the warehouse made by
one or multiple storage zones), the decision-maker imports the location coordi-nates for each zone and arranges them
according to the overall warehouse layout configuration.
The DSS calculates, for every location, the forward area and reserves the travel path (in terms of distances) from the
shipping and receiving docks and those from/to each other (see Fig. 8b).
The what-if simulation analysis involves inbound (e.g., put-away, restocking) and outbound (e.g., order picking)
operations and provides a useful tool to assess system performances, including costs (i.e., in terms of the travelled distance
and time) within a specific horizon of analysis. The DSS reports a complete panel of statistics and KPIs to evaluate the
efficacy and efficiency of the layout, allocation and assignment configuration. A list of statistics includes the travelled
distance (horizontal and vertical) and time due to pick-path, travelled distance (horizontal and vertical) due to put away and
replenishment, time waste due to stock-out, number of replenishments per each SKU, number of visited aisles, as a metric
of the vehicles congestions, spatial pick-density, and other aspects.

4. Case study

In this section, the proposed DSS has been applied for the design and performance assessment of a real-world
warehousing system. In particular, this case study addresses a spare parts management system for an international brand of
the automotive industry. A logistic firm operating worldwide provides the logistics services of transportation (inbound and
outbound) and warehousing for an important automotive company to supply the demand of spare parts to hundreds of
Italian dealers. This system is a regional distribution centre (RDC) that accounts approximately 8000 SKUs as spare parts,
ranging from bonnets to screws. The high variability of SKUs in size, weight, and shape is typical for automotive industry,
and requires proper storage management practices. The analysed system accounts twenty-four aisles arranged in a multi-
zones warehouse, which holds four storage zones grouping homogeneous SKUs in size and shape of unit load and similar
in weight. Each storage area presents a different type of racks, which is suitable to allocate a specific set of SKUs (e.g.,
cantilever for door, bin shelving for air-filter, etc.).

The observed low-level single-order picker-to-part and forward-reserve OPS represents a relevant sample to assess the
effectiveness of the proposed tool. The receiving and shipping activities are decoupled and limited respectively to the left
and right of the dockside. Although the receiving and shipping docks are distributed along the warehouse side, the picking
process starts on the bottom left corner of the system and the ends on right bottom corner. These two control points
represent respectively the parking of walkie-stackers and roll-containers and the sorting/packing station for the orders to be
shipped. The low-level storage area (i.e., the forward area) is 25,000 square metre wide. Customer orders, made by many
order lines, accounts on average 37 lines, results in long time-effective picking missions, since the pickers has to achieve in
sequence products which are far located one from the others. The presence of narrow aisles does not allow reverse back,
and the ‘‘traversal’’ visiting strategy is adopted either in real world than in the simulation.

The complexity of the system bases on the disomogeneity of both the SKUs and the processes. The increasing complexity
of modern supply chain shifts the role of warehousing systems in addressing demand variability, pushing logistic providers
to handle both homogeneous and heterogeneous flows. Such a trend results for the observed warehouse in articulated
inbound/ outbound operations, which include the truck unloading, the check of loads, put-away and replenishment, and the
order picking. A dashboard of KPIs involving put away, replenishment and order picking missions allows the decision-
maker to address operative criticalities and propose strategies for both layout re-design and operations improvements

The DSS implements a what-if multi-scenario simulation to compare how different allocation and assignment strategies
affect the performance of the inbound/outbound operations in the observed warehousing system. The simulation analysis
does not involve layout leverage, which is common to any proposed scenarios, since the client had no budget for layout re-
design or infrastructure refurbishing. For sake of brevity, the comparative analysis regards just with the reduction of travel
distances, as an approximation of the operative time.

Table 2 illustrates the results of a simulation campaign conducted on the historical set of inbound/outbound annual
operations, which account about 970,100 picking lines, and 25,500 replenishment missions. The colours yellow, red, and
blue refers respectively to the layout, allocation and assignment steps of analysis implemented into the DSS (see Fig. 2).

The overall warehousing system is composed by the four storage areas, which are independently designed in
accordance with the allocation and assignment policies. In particular, the warehouse scenarios are organised through the
adoption of the following rules and parameters:
3 allocation strategies (i.e., EQS, EQT, OPT).
4 assignment strategies (i.e., popularity, COI, turn, OC). 1 visiting strategy (i.e., traversal).
1 routing heuristic strategy (i.e., nearest neighbour).

Different allocation strategies result, first, in different values of total replenishments within the observed horizon of
time, and second, if combined with different assignment strategies, in influencing the location of the SKUs in the forward
area, thereby affecting the total travelled distance for all inbound/outbound operations. The illustration of the alternative
warehouse scenario allows the decision-maker to recognise the influence of decisions on SKU allocations and assignments
in both the forward and reserve storage areas. The saving of replenishment missions occurred by EQT and OPT strategies,
are not enough to justify their implementation considering the overall costs. Indeed, the combi-nation of an EQS strategy
and popularity rule accomplishes reducing the total travelled distance primarily because of the picking activities.

This section gives a picture of the potential multi-leverage analyses conducted through the proposed DSS. Significant
time-savings can be generated by a re-allocation and re-assignment of SKUs within the forward area of a multiple zones
warehousing system. In the following section, the potential applications of the proposed DSS will be described with the
focus on the enhancement opportunities in tackling real world instance and both strategic and operative warehousing
decisions.

5. Discussion

Despite of the increasing trend of lean paradigm in production and distribution operations, warehousing systems are
still neces-sary to address the demand variability and seasonality, to match vendors and consumers in global trade, to hold
products and sustain the customer service level. The reduction in demanded quantity joined by the customization of items,
results in raising the complexity of the warehousing operations, which are called to achieve high performances and to make
goods travelling fast throughout the distribution pipeline.

The proposed DSS supports the decision-maker in addressing warehouse operations, which are highly dependent by a
broad set of factors including the layout, the storage equipment and infrastructure, the set of SKUs,the order profile, the
SKUs turnover, the routing policies for put away and picking missions, the goal performances in terms of time efficiency,
space efficiency or both.
The implementation of this tool for real-world instance has different purposes. First, it supports the decision-maker in
handling long-term strategic decisions, based on the estimation of requirements of space and investments (i.e., costs for
racks and storage equipments) necessary to arrange a new storage area from green-field. Second, it addresses mid-term
tactical decisions, involving the definition of the storage areas devoted to picking rather than bulk storage (i.e., forward-
reverse low-level system vs. high-level system), the size and shape of each storage area, and the opportunity to set multiple
storage areas dedicated to different classes of SKUs. Tactical decisions include also the analysis of the appropriate storage
quantity to allocate to each SKU, thereby affecting the reorder quantity from the distribution nodes at previous stage of the
supply chain. Third, the DSS handles operative short-term decisions, based on the assignment strategies of SKUs to
locations, the definition of the best performing routing policy, the selection of the retrieving strategy to adopt (i.e., single-
order vs. order-batching). The what-if multi-scenario simulation analysis assesses the operative performances of each
scenario, providing improvements solutions and enhancement guidelines with operative, tactical and strategic horizons of
analysis.

Based on the described functionalities, it provides various levels of assistance to different users. Specifically, the DSS
supports the third part logistic (3PL) managers in facing daily concerns on the management of multiple-client storage
systems, characterised by high-variability in items, storage racks, and turnovers. The tool depicts a detailed dashboard of
the operative performances of a generic storage area (i.e., a generic client), with suggestion for tactical and operational
improvements and tips for scheduling labour among different areas. Furthermore, the DSS offers to warehouse owners the
opportunity to simulate the operative savings (i.e., time, costs, and space) achieved by the combination of allocation and
assignment strategies, which results in changing approaches for the management of SKUs slotting. Finally, the DSS allows
researchers approaching different real case studies, testing the effectiveness of models and heuristics on providing
performing solutions and creating knowledge over the most critical and recurrent storage issues.

Table 3 reports the obtained results by the implementation of the DDS with three real case studies. These profiles were
selected as the basis for system validation since they were most representative for computational complexity and robustness
of enterprise datasets. Systemic analysis of the three profiles identifies major opportunities for improvement over the AS-IS
scenario. The three profiles differ for the industrial sector of application, the set of SKUs, the purpose of the client and the
related implemented analyses.
 Despite of the observed business, the DSS tends to support the grouping of the fast-moving SKUs, grouped per affinity
(i.e., Case 1), per popularity or cube-per-order index (i.e., Case 2 and Case 3), within the most convenient storage areas,
thereby eliciting the need of dedicated golden areas, potentially supported by different types of technologies (i.e., conveyor,
semi-automated storage/ retrieving systems). The tool properly arranges the available storage space for both forward-
reserve and high-level storage systems, involving the design of different storage areas, which increases the pick density and
space efficiency of the system.

6. Conclusions

An original decision-support system for picker-to-part storage system design and operations management is illustrated.
The proposed DSS comprises a user-friendly tool for supporting practitioners, managers, decision-makers, and logistics
providers by addressing real case studies and experimental analyses over the design and operations control of the storage
systems. This tool enables us to gather and store information from enterprise WMSs and to elaborate, through an efficient
DBMS architecture, a set of data-oriented design solutions and configurations. The tool aims to design multi-zone storage
systems and implements a wide panel of algorithms and methods that address different stages of analysis (e.g., storage
allocation, assignment, batching, zoning, routing). Results and statistics on performances and costs due to a generic
warehouse scenario are computed through a what-if simulation analysis. An implemented graphic interface draws two-
dimen-sional and three-dimensional views of the designed storage scenario, adopting real commercial rack components
with the purpose of providing a ready-to-print release of the warehouse for logistic providers and engineers.

Further developments are expected on the implementation of innovative methods, models and algorithms, to address
warehouse layout, storage allocation and storage assignment issues in the presence of automated storage solutions and
equipment for part-to-picker systems (e.g., automated storage and retrieval systems (AS/RS), mini-load, carousels).

A useful module, integrating a cam interface for barcode reading, could be implemented to support the introduction and
registration of new SKUs and the updating of the enterprise SKU master file. This functionality might respond to the
problem of periodical and partial storage rearrangement rather than overall warehouse redesigning.

The educational purpose of this work is to provide a set of flexible interactive instruments to create and disseminate
knowledge among logistic providers, practitioners, and managers, and to improve industrial engineers’ backgrounds and
expertise over the most critical storage issues. Lastly, the designed tool, similar to any other computer-aided system,
attempts to support, but not replace, the decision-maker, who responds daily to strategic design and operations management
within a storage system.