Chapter 16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System: A Case Study Alberto Regattieri, Riccardo Manzini and Mauro Gamberi
Abstract The logistics reengineering process (LRP) is a useful industrial engineering and management technique for achieving significant improvements in operational efficiencies for products quality services in a warehouse/order fulfillment system. In warehousing systems the picking process usually has a significant impact on logistic performance, customer service levels and costs, hence improvement activities are attractive and important. This chapter presents the application of an LRP process in an Italian distribution company, which is a distributor of home furnishings and health care products. In particular, the proposed optimization process is focused on the Order Fulfillment Process (OFP). The main aim of this chapter is to present a methodology to make an effective analysis of an OFP system and, mainly, to present the results, opportunities and criticalities arising from its application. The benefits are significant both in terms of traveled distance savings and manpower usage reduction. These results demonstrate that ‘‘soft’’ reengineering improvements can significantly affect processes, procedures, rules and strategies, can reduce logistics costs and improve customer service levels without introducing ‘‘hard’’ improvements and system modifications, e.g. new equipment, personnel, and machinery.
A. Regattieri (&) R. Manzini DIEM—Department of Industrial and Mechanical Plants, Bologna University, v.le Risorgimento 2, 40136 Bologna, Italy e-mail:
[email protected] M. Gamberi Department of Management and Engineering, DTG University, Padova, Italy
R. Manzini (ed.), Warehousing in the Global Supply Chain, DOI: 10.1007/978-1-4471-2274-6_16, Springer-Verlag London Limited 2012
425
426
A. Regattieri et al.
16.1 Introduction During recent decades the Logistics Reengineering Process (LRP) has been used by many companies in the pursuit of efficiency through technology and organizational developments. In manufacturing warehouses the picking systems usually have a significant impact in terms of logistic process, customer service levels and costs; hence improvement activities are attractive and important. Ozcelik (2010) and Chien-wen (2007) examine whether the implementation of logistics process reengineering projects improves company performance by analyzing a comprehensive set of data on firms in the United States and Taiwan, respectively. Moreover, a series of studies in the early 1990s found that a significant number of LRP initiatives failed or delivered less than they promised (Chien-wen 2007). LRP process is potentially very attractive but is not always easy. This chapter presents the application of an LRP process in an Italian distribution company, named Alpha (a disguised name), which sells home furnishings and health care products. The materials received from the suppliers are checked, accepted and then stored in a warehouse system. The customer orders generate intensive split case picking activities and finally the materials are shipped out. To satisfy and respond quickly to customer demand, Alpha is now strongly focusing on warehouse management and on the order fulfillment process with the aim of strengthening its competitiveness. Alpha is engaged in activities ranging from the supply of input materials to the production and delivery of products, in an attempt to obtain the best added value while reducing the entire supply chain cost (Croom et al. 2000; Piramuthu 2005). In general there is a large amount of complete literature dealing with Business Process Reengineering (BPR), but there are very few sources that explicitly discuss this process applied to the Order Fulfillment System. In general the Order Fulfillment Process (OFP) has been recognized as one of the core business processes in any organization (Kritchanchai and MacCarthy 1999; Turner et al. 2002; Manzini et al. 2005, 2007; Park and Lee 2007). An OFP starts with receiving customer orders and ends with delivering products. It includes activities such as purchase orders (PO) planning, customer order processing, stock checking, order picking activities, final shipping assembly and delivery. While in different companies, OFPs are executed differently in accordance with their unique business characteristics (e.g. product characteristics and lead times), they fulfill two common objectives (De Koster et al. 2007; Ozcelik, 2010): 1. The delivery of products to satisfy customer expectations at the right time, right place, and right quantity. 2. Effective management of the uncertainties from internal and external environments (delay in supply, quality problems, etc.). In any OFP, the wide range of activities involved is carried out by operators from different functional units (e.g. sales personnel for processing customer orders and logistics personnel to manage picking and packing activities). As a result, an
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
427
OFP is complicated and requires activity integration and coordination. In this condition, it is necessary for companies to reengineer their OFPs so that they can be better integrated into the corresponding supply chains (Kritchanchai and MacCarthy 1999; Turner et al. 2002; Waller et al. 1995). Furthermore, logistics systems are dynamic, which leads to the importance for companies to reengineer their OFPs systematically to exploit saving and efficiency. The studies in the literature focusing on Order Fulfillment Systems usually pay attention to success factors such as impetus, opportunities and strategies, but do not address the operational issues (Grover Kettinger 2000; Kleiner 2000). Others studies deal with several particular problems such as order batching, picking area design and product customization (Bartholdi and Hackman 2010; Yu and de Koster 2009; de Koster et al. 2007; Zhang et al. 2010). In view of these limitations this chapter seeks to present a case study dealing with a practical investigation of an OFP system. The final aim is to propose a methodology to make an effective analysis of an OFP system and, mainly, to present the results, opportunities and criticalities arising from its application. This chapter, by means of a case study, discusses original and innovative approaches for reengineering Order Fulfillment Systems based on new material allocation strategies and product clustering. The remainder of this chapter is organized as follows: Sect. 16.2 presents the logistic chain of the company (subject of interest as a case study). Section 16.3 presents the results of the AS-IS analysis; the main collected criticalities are discussed in Sect. 16.4. Finally, several reengineering activities are presented in Sect. 16.5.
16.2 Alpha Company Logistic Chain Alpha sells home furnishings and health care products. Its supply chain involves a great number of suppliers and customers (i.e. hundreds). Figure 16.1 shows the activities related to the order fulfillment system, normally placed along the supply chain. The gray box lists the logistics activities directly connected with material handling. Reengineering allows breakthrough improvements by discarding existing malfunctioning processes. This effort may not be rewarded without a systematic analysis of the AS-IS condition. The improvement activities must be developed considering the results of this assessment. At Alpha the AS-IS analysis concentrates on the material handling process, allowing several steps, in particular: • Layout analysis The aim is to study the layout to locate the warehouse area, the other areas (e.g. receiving area, order setup area, etc.) and their characteristics; • Analysis of material receiving procedures and activities This part focuses on the investigation of procedures connected with the
428
A. Regattieri et al.
acceptance and inspection of material from suppliers waiting in the checking area. The main aspects considered are workloads, (i.e. personnel and shifting), the equipment used and the activities performed (in term of targets, durations and information technology support needed); The approach mentioned in the last bullet point is repeated to study the other activities, in particular in the following steps: • Analysis of the storage process This process deals with the handling activities performed on the accepted materials from the check-in area to the storage area; • Analysis of the picking and refilling processes The picking and refilling processes are the core logistics tasks executed. In these phases the goods are collected considering customer orders and the picking areas are replenished. The relationship between the materials in the stock area and in the picking area plays a crucial role in the performance of the entire order fulfillment system. • Analysis of order setting and shipment processes. The materials and related orders are checked and then the checked goods are packed and forwarded to the shipment process and finally delivered to customers. The AS-IS analysis identifies a set of main criticalities and of further development areas that are the starting point for the following improvement activities. The following section presents the entire AS-IS analysis and some examples of improvement activities on the order picking process.
16.2.1 Definition and Notation of Symbols ASj CLi fi IASj ICj LRP OP OPi OPTi PDA Rj RTc RTf SKU Vi
available space in cluster j products cluster i rate of material flow through the warehouse for the item i available space for an item of cluster j number of items in cluster j Logistics reengineering process Order picking operator i optimal fraction of available space devoted to item i Personal digital assistant residuals for day j, between forecasted and collected discharging time discharging time collected for day j discharging time forecasted for day j Stock keeping units volume of product i
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
Wi WMS
429
weight of product i Warehouse Management System
16.3 AS-IS Analysis 16.3.1 Layout Figure 16.2 shows the layout of Alpha. Excluding the office building, the total area is about 10,000 m2, of which roughly 5,000 m2 are devoted to storage. The shelves are equipped with pallet racks that are organized into 13 aisles. The racks have six levels: the ground level is dedicated to picking activities; levels 1–5 contain the Stock Keeping Units (SKUs). The forward-reserve picker to part policy (Caron et. al. 2000) is adopted. Lower rack levels are used for manual Order picking (OP) (the forward area), while higher levels contain bulk storage (the reserve area). The SKU storage capacity is about 9,200. The SKU used as a reference is the pallet EUR-EPAL 800 9 1200 mm. A secondary storage area is available, about 480 m2, where out of shape materials are located, which have larger external dimensions than the racks. The building presents 12 truck docks, seven used for shipping material and five used for receiving material. Between the truck docks and shelves two depot areas, 450 and 700 m2 respectively, allow temporary storage of goods. Materials arriving (from suppliers) are discharged and placed in a temporary receiving area. The check-in procedure requires a physical check and data entry in the Enterprise Resource Planning System (ERP). When this procedure is complete the storage activities are executed. The shipment area collects the material waiting for delivery to customers. In this area packing activities are carried out, when necessary. Considering the retrieval activities, picking is a Less than unit load with multiple stops per trip. The pickers retrieve sets of items or multiple handling units of the same item on a single Order Picking cycle (Manzini et al. 2007). Each picker usually picks a complete customer order during a mission. All the activities are supported by two software packages: an ERP that manages the entire supply chain from customer orders to customer invoicing (including supplier orders) and a Warehouse Management System (WMS), which focuses on supporting the warehouse activities (e.g. material acceptance, labeling, picking and order pack listing). In the following paragraphs all the logistics activities represented in the gray box in Fig. 16.1 are discussed with the same approach, such as the analysis of the workload involved and the equipment used, the description of activities by means of a flowchart and finally the analysis of statistics and observations.
430
A. Regattieri et al.
Fig. 16.1 Alpha company supply chain (order fulfillment system detail)
Alpha activities C orders
orders generation
forecast
Selection of suppliers Negotiation with suppliers Issuing & purchasing Order paying
Reception of materials
CUSTOMERS
SUPPLIERS
S orders
Inspecting quantity & quality
Storaging materials Material picking & refilling Material Shipping
Invoicing
16.3.2 Acceptance of Materials Four operators are normally employed in the receiving area (named Op1, …, Op4). Their shifts during the week are organized and shown in Table 16.1. Their fleet of vehicles consists of five pallet trucks, 1,200 kg load capacity, 2,390 mm Duplex mast legs and a forklift truck with load capacity of 1,600 kg, Triple mast legs. The workload in the receiving area is concentrated in the morning. This is a normal condition due to the use of industrial vehicles for transporting the materials both from suppliers and to customers.
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
431
Table 16.1 Workload in receiving area 6 6 6 6
6 AM–1.45 PM 6 AM–1.45 PM 12 AM–7.45 PM 6 AM–1.45 PM
AM–1.45 AM–1.45 AM–1.45 AM–1.45
Wednesday PM PM PM PM
6 6 6 6
AM–1.45 AM–1.45 AM–1.45 AM–1.45
Thursday PM PM PM PM
6 6 6 6
AM–1.45 AM–1.45 AM–1.45 AM–1.45
6 AM–1.45 PM 12 AM–7.45 PM 6 AM–1.45 PM 6 AM–1.45 PM
U. S.
Hm=7.75
Friday PM PM PM PM
1.30 2.15
Tuesday
Op1 Op2 Op3 Op4
2,50 4,20
Operators Monday
2.50 3.50
1
2.50 3.50
2 3 4
order setup area & shipment area
5
2.50 3.50
2.50 3.50
2.50 3.50
2.50 3.50
6
docks
2.50 3.50
2.50 3.50
7
2.50 3.50
pallet racking area
8 2.50 3.50
9 2.50 3.50
10 11
receiving area
2.50 3.50
2.50 3.50
12 13
out of shape storage area
recharging forklift area
office
Fig. 16.2 Layout of Alpha company
The operators involved in receiving are responsible for the activities of vehicle hauling in a dock to the material depot in the receiving area. These material handling activities are supported by several operations performed in the ERP system and in the WMS system. Figure 16.4 shows a flowchart describing all these tasks. Different kinds of industrial vehicles arrive at Alpha. Figure 16.3 shows the average percentage collected over a 3 month assessment.
432
A. Regattieri et al.
Fig. 16.3 Percentage of different kinds of industrial vehicles arriving to docks
van 17%
lorry trailer 19%
truck 23% tractor trailer 41%
Figure 16.5 presents the average durations, collected during the assessment, for carrying out the entire procedure depicted in Fig. 16.4, from vehicle docking to vehicle departure and item labeling. Different kinds of vehicles have different load capacities hence the time taken is quite different. Another interesting observation emerges from the analysis of the number of packages unloaded during the assessment period (i.e. in an SKU there are usually several packages), which vary according to the different days of the week. Monday and Friday are definitely more critical compared to the other days, in terms of the amount of material to be unloaded. This typical behavior asks for the introduction of variable workloads but there is currently a fixed number of operators for this task. Figure 16.6 shows the average number of packages unloaded during the week and Fig. 16.7 presents the corresponding analysis of variance. In terms of analysis of variance the average number of packages processed per day is 9,781, but the spectrum is wide as shown in the Figure. Over recent years, Alpha has developed a model based on the concept of standard unloading times for each different transport system, in an attempt to schedule the arrivals of the supplier vehicles with the aim of balancing the workload in the goods receipt process. This model appears to be insufficiently accurate. Figure 16.8 shows the analysis of residuals Rj, (minutes) for day j, as the result of comparison between the forecasted and the collected daily discharging time values. Rj ¼ RTf RTc with: RTf discharging time forecasted for day j (min); RTc discharging time collected for day j (min);
ð16:1Þ
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
Fig. 16.4 Receiving activities flowchart
433
Truck docking Packing list delivering (from truck driver)
SKU unload no
Equal to previous SKU? yes
SKU bar code reading Packagesbar code reading Single itemsbar code reading Comparison packing list/order in WMS system
picking and reserve area SKU handling in receiving area
Another SKU to unload?
yes
no
Empty SKUs loading on truck (if needed) Document ruling (to truck driver)
Truck leaving
Packing list data entry in WMS system Exporting data in ERP system Labels printing and items labeling stop
Fig. 16.5 Receiving average times and average value of SKUs unloaded by kind of vehicle
receiving time (min)
SKUs unloaded (average)
87.4 72.6 53.2 36.7
43.7 30.7 10.1
lorry trailer
tractor -trailer
truck
2.5 van
434 Fig. 16.6 Average number of packages processed per day in receiving area
A. Regattieri et al. packages/day 17469 15531
8961
Monday
5914
6247
Tuesday
Wednesday
Thursday
Friday
Fig. 16.7 Number of average packages received per day: analysis of variance
Fig. 16.8 Discharging time model accuracy: analysis of variance of residuals Rj
In conclusion, the scheduling of vehicles arriving is not accurate and the effect of the beginning and the end of week due to the commercial requirements of suppliers and freight forwarders is significant. This results in serious inefficiencies and extra costs (i.e. work overtime, traffic congestion, equipment damage and industrial accidents).
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
435
Table 16.2 Workload in storage area Operators
Monday
Tuesday
Wednesday
Thursday
Friday
Op5 Op6 Op7 Op8 Op9 Op10
12 AM–7.45 PM 6 AM–1.45 PM 12 AM–7.45 PM 6 AM–1.45 PM 12 AM–7.45 PM 6 AM–1.45 PM
6 AM–1.45 PM 12 AM–7.45 PM 12 AM–7.45 PM 12 AM–7.45 PM 6 AM–1.45 PM 6 AM–1.45 PM
12 AM–7.45 PM 6 AM–1.45 PM 6 AM–1.45 PM 12 AM–7.45 PM 12 AM–7.45 PM 6 AM–1.45 PM
6 AM–1.45 PM 12 AM–7.45 PM 6 AM–1.45 PM 6 AM–1.45 PM 6 AM–1.45 PM 12 AM–7.45 PM
6 AM–1.45 PM 6 AM–1.45 PM 6 AM–1.45 PM 6 AM–1.45 PM 6 AM–1.45 PM 12 AM–7.45 PM
Fig. 16.9 Storage activities flowchart
Receiving area empty ?
yes
stop
no
SKU bar code reading by PDA Handling to storage location suggested by WMS
Storage location eligible ?
yes
Location bar code reading by PDA
Searching of an eligible storage location
New location data entering in WMS
SKU stocking
16.3.3 Storage After the acceptance procedure (para 3.2) the materials wait in the receiving area until operators carry them into the storage area (i.e. reserve area). This operation is usually performed by six operators (named Op5, …, Op10). Their shifts during the week are organized and shown in Table 16.2. Their fleet of vehicles consists of 6 reach trucks, 1,600 kg load capacity and Triple mast legs. In all the storage activities operators are supported by the WMS system by means of personal digital assistant (PDA) devices. PDA operators know the assigned storage location, read bar codes and, if needed, can edit incorrect information on the WMS. Figure 16.9 presents a flowchart dealing with storage activities.
436
A. Regattieri et al.
Fig. 16.10 Time sampling of storage activities
Class of times Table 16.3 Product clusters assignment in picking area
Aisle
Assigned clusters of products
1 2 3 4 5 6 7 8 9 10 11 12 13 Out of shape
CL1, CL2 CL3 CL4 CL5, CL6 CL7 CL8 CL9, CL10, CL11, CL12 CL13, CL14, CL15 CL16, CL17, CL18, CL19 CL20 CL21 CL22, CL23 CL24 CL25
During the assessment much of the sampling is focused on the duration of the storage activities revealing an average time of about 6.59 min from the SKU pick up operation in the receiving area to the SKU drop off operation in the storage location. Figure 16.10 goes into more depth on the analysis of variance of the storage time. The storage locations of products (reserve area) are related to their picking locations (forward area). Both positions are managed by the WMS system by means of product clustering and dedicated aisles. Alpha manages about 7,150 different products. They are grouped into 25 clusters (named CL1, …, CL25) stratified by volume, dimensions and weight. In the picking area a single cluster or group of clusters is assigned to storage slots pertaining to racks looking towards the same aisle. The reserves of products in the storage area are located in higher levels as near as possible to the corresponding
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
IASj
Fig. 16.11 Available space for an item of cluster j (IASj) in reserve and in forward area
15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
437
IASj reserve area (SKUs locations) IASj forward area (picking locations)
1
2
3
4
5
6
7
8
9
10 11 12 13
Aisle
picking location on the lower level. This procedure is entirely supported by the WMS using PDA devices. Table 16.3 shows the product clusters assigned to different aisles. The allocation of clusters/aisles appears to be slightly off balance in terms of the space required and the allocated space, both in the picking area and the reserve area. Consider IASj as the available space for an item of cluster j in terms of SKU storage locations in the reserve area and in terms of allocated picking space in the forward area: IASj ¼
ASj ICj
ð16:2Þ
with: available space in cluster j (SKUs locations in the reserve area or picking ASj locations in the forward area); ICj number of items in cluster j IASj has a wide range and is dispersed as clearly shown in Fig. 16.11 Alpha applies different supply policies to different products. Furthermore, the products have very different characteristics in terms of shape and dimensions. In conclusion clusters need a different amount of space both in the reserve area and the forward area. But these differences do not explain the wide range of variation in the IASj parameter. Furthermore several values of IASj are less then the unit. This behavior results in the continuous use of the manual location of materials by operators in an aisle not suggested by the WMS. In these conditions picking traveling time and distances clearly increase, along with the number of refilling missions (restocks).
16.3.4 Restocking Process The order fulfillment activities collect the products in the picking locations, then refilling, i.e. restocking, is necessary. This restocking task is done by the same
438
A. Regattieri et al. Reading on PDA of 1° item in restockslist Travelling in storage location and SKU picking Scanning of item’s bar code and of storage location’s bar code
Travelling in Picking location and forwardarea refilling
Scanning of picking location’s bar code and dropingoff confirmation no
no
Restocks list empty ?
On hand materials ?
yes
yes
stop
Original stock location is ok? (by operator)
no
Selection of a new location in storage area (by operator)
yes
Travelling in storage location and SKU droping off Scanning of storagelocation’s bar code and droping off confirmation
Fig. 16.12 Refilling of picking area activities
personnel who stocks the materials (i.e. Table 16.2). WMS, by means of PDA, guides the operators to get restocks following a list that ranks product in hand order and products which have the empty locations in the picking area. The sampling activities carried out during the assessment have highlighted two main characteristics regarding the restock process: the average duration of a restock mission is about 8’55’’ min and the average distance travelled is about 34.0 m. Figure 16.12 presents a flowchart describing the refilling of picking area activities.
16.3.5 Order Picking Orders fulfillment is the core activity in Alpha. Its importance emerges due to strongly differentiated products with shorter life cycles, low volume and low customer delivery time accepted that Alpha normally manage.
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
439
Table 16.4 Workload in storage area Operators
Monday
Op11 Op12 Op13 …. Op25
7 7 6 6 6
AM–2.45 AM–2.45 AM–1.45 AM–1.45 AM–1.45
Tuesday PM PM PM PM PM
7 7 6 6 6
AM–2.45 AM–2.45 AM–1.45 AM–1.45 AM–1.45
Wednesday PM PM PM PM PM
7 7 6 6 6
AM–2.45 AM–2.45 AM–2.45 AM–2.45 AM–2.45
Thursday PM PM PM PM PM
7 7 6 6 6
AM–2.45 AM–2.45 AM–1.45 AM–1.45 AM–1.45
Friday PM PM PM PM PM
7 7 6 6 6
AM–2.45 AM–2.45 AM–1.45 AM–1.45 AM–1.45
PM PM PM PM PM
Retrieval activities can be defined as less than unit load systems with multiple stops per trip. Order pickers retrieve sets of items or multiple SKUs of the same item on a single order picking cycle. They visit different slots in the forward area before going to the shipment area. Each picker is responsible for picking a complete single customer order during a mission. The picking operation is usually performed by 6 operators (named Op11, …, Op25). Their shifts during the week are organized and shown in Table 16.4. Their fleet of vehicles consists of 15 order picker trucks, 2,000 kg load capacity, 2,350 mm legs. Figure 16.13 shows the flow of activities to perform a customer order picking cycle. The starting point is the delivery of the customer order list to the picker by the dispatch department. The pickers are supported by WMS through a PDA system including voice technology and speech recognition to communicate. In particular, the pickers use headsets and microphones to receive instructions by voice, and verbally confirm their actions back to the system. As mentioned before the adopted storage location assignment policy is based on cluster allocation in defined aisles. The products are grouped into clusters according to their characteristics (i.e. volume, weight, dimensions, etc.), and the clusters are assigned to different aisles as reported in Table 16.3. Both the cluster assignment to aisles and the space allocation of the picking area to different products are suggested by the WMS, exclusively dictated by empirical evaluation and the working experience of the warehouse managers. During the assessment the characteristics of customer orders are analyzed, in particular the number of different lines per order and the corresponding volume of material. Each order line implies a trip to reach the appropriate slot in the forward area and several retrieval activities on single or multiple packages of product. The items marketed by Alpha are very different in terms of weight and size, hence the orders have very different volumes in terms of space used. The capacity of the order picker truck can almost always contain the volume of material corresponding to a customer order. The average number of lines per order is 83.7 and the average order volume is 1424 dm3, but the values are very dispersed. Figures 16.14 and 16.15 show the statistical analysis of these two parameters for orders collected during the period December 2008–December 2009.
440
A. Regattieri et al. Collection of a new order list Order code reading (PDA) and confirming(voice)
Target aisle and location communicating (voice)
Target aisle and location travelling
Target reached confirming (voice)
Product quantity (to pick) communicating (voice)
Forward product quantity sufficient ?
no
yes
No pickup communicatingand confirming (voice)
Packages picking up
Picking up confirming (voice)
End of order?
no
yes
Travelling to shipping area
Order packing
Fig. 16.13 Picking activities flowchart
During the assessment the productivity of the pickers is evaluated both by means of tests carried out in the field and statistical tracing of WMS data. The average order picking mission takes 39.2 min and includes about 83.1 order lines corresponding to 118.2 packages. The high number of lines per order in comparison to the number of aisles results in a complete crossing of the warehouse area for each customer order fulfillment. In other words during an order mission a picker visits all the aisles typically with a traversal policy (Manzini et al. 2005; Bindi et al. 2009). The unbalanced product
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
441
Fig. 16.14 Lines per customer order (Dec 2008–Dec 2009)
Fig. 16.15 Volume per customer order (Dec 2008–Dec 2009)
clustering and aisle assignment, as revealed by IASj analysis (Fig. 16.11), cause an unbalanced use of slots and aisles. The analysis of popularity of products (i.e. the number of requests for a given products, so the number of times a picker must travel to a storage location for a given product) related to the spatial position in the layout of warehouse reveals a critical situation typically resulting in the over-crowding of trucks and a high number of restocks. Figure 16.16 shows the popularity of products in the forward area along different spatial coordinates of warehouse. The diameter of the circle is proportioned to the popularity value.
16.3.6 Orders Shipment Setting The pickers leave the material collected in the shipping area and several activities to prepare this material for the final loading into the industrial vehicle must be
442
A. Regattieri et al.
Fig. 16.16 Analysis of popularity of product versus spatial coordinates
performed. The load capacity of a vehicle includes material corresponding to several customer orders, typically from two to ten orders. These operations are usually performed by four operators (named Op26, …, Op29). Their shifts during the week are organized and shown in Table 16.5. Their fleet of vehicles consists of one reach truck, 1,600 kg load capacity, triple mast legs and three robopacks to apply the plastic film to the pallets. Normally before picking operations, the orders are clustered by the dispatch department into different groups considering the customer locations, the volume of materials and the kind of available vehicles for the considered delivery date. Considering a single cluster of orders, the analysis of customer material acceptance times (often there are significant constraints) and the best vehicle routing (also taking into consideration traffic congestion) give the pallet priority in vehicle loading activities.
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
443
Table 16.5 Workload in shipping area Operators Monday Op26 Op27 Op28 Op29
6 AM–13.45 12 AM–7.45 12 AM–7.45 12 AM–7.45
Tuesday PM PM PM PM
Wednesday
6 AM–13.45 12 AM–7.45 12 AM–7.45 12 AM–7.45
Fig. 16.17 Orders shipment setting flowchart
PM PM PM PM
6 AM–13.45 12 AM–7.45 12 AM–7.45 12 AM–7.45
Thursday PM PM PM PM
6 AM–13.45 12 AM–7.45 12 AM–7.45 12 AM–7.45
Friday PM PM PM PM
6 AM–13.45 12 AM–7.45 12 AM–7.45 12 AM–7.45
PM PM PM PM
Gathering of arrival time and route plan of vehicle
Analysis of assignment Vehicle –Group of orders (by dispatching office)
Design of pallet loading order on vehicle (based on vehicle routing)
Pallet packing (plastic film) and labeling
Pallet loading on vehicle
no
Group of orders completed? yes
Packing list delivering (todispatching office)
Vehicledeparture
A customer order is usually split into different pallets (EUR-EPAL 1,200 9 800 mm). At the end, the operators hand over the final packing list to the dispatch department, which performs the final check and allows the truck to depart. Figure 16.17 shows the flowchart of typical order shipment setting activities. During the assessment several sampling activities on the time duration of activities were carried out. The reference must be the pallet because the orders are split into a variable number. The average times are: pallet setting before applying the plastic film 2.3 min, film application 1.5 min and pallet handling to vehicle 2.5 min.
16.4 Main Criticalities The AS-IS analysis is proposed to identify the critical processes. Process-flow diagrams, process analysis worksheets and data summary charts are prepared as effective tools to get a thorough understanding of the existing process with a view
444
A. Regattieri et al.
to focus on the possible areas of improvement. In particular procedures, workflows, and resource consumption in terms of workforce, equipment and information technology support are investigated and discussed. In the acceptance process the number of packages processed per day is different on different days of the week. Typically Monday and Friday are overloaded considering that the workforce is constant throughout the week. The final result is significant mental stress for operators and need to work overtime and hence extra costs. To provide the time consumption in the SKU unloading process, Alpha uses a model based on standard time for each kind of industrial vehicle. Experimental evidence shows that the accuracy is not sufficient and the model does not represent valid support for the vehicle arrival planning process. Another important criticality deals with the space allocation policy used both for product clusters and for single products in a single cluster. Alpha adopted a fixed assignment aisle-cluster of products in the picking area. One or more clusters are dedicated to racks corresponding to a single aisle in the picking area. The products in the reserve area are allocated to higher levels of racks as near as possible to the position of the same products in the picking area (lower level). When picking area is empty product in reserve area are pulled down usually by forklift. This assignment aisle-product clusters is managed by WMS but only considering the experience of the managers that force materials in positions which are different in comparison to WMS suggestions. Their experience also guides the allocation of the space of racks for products into a single cluster. Considering the constraint of the total space for the cluster or the group of clusters (i.e. the assigned aisle of racks) managers divide the racks into four kinds of slots which have different capacities, ranging from one to four single units of slots, and assign different products to these different slots. This empirical double assignment (rack aisles-product clusters and slotsproducts) and the following allocation of material in the reserve area lead to several problems. The products have very different sales levels in term of quantity but similar space is allocated in the picking area, so order pickers often do not find sufficient product quantity to fulfill the order (and temporarily skip the item) or pick up all the product stock in the picking area. In both cases a high number of refilling missions (restocks) are needed. This unbalanced space assignment, also revealed by the popularity analysis, generates significant traffic congestion in aisles 6–11. To limit this problem, pickers often force manual allocation of materials in different slots from the positions suggested by the WMS. This bad procedure leads to long distances between the same material in the picking area (lower level) and the reserve area (higher levels) increasing the costs of the refilling missions. These kinds of missions require long distances in terms of space because the allocated space in the reserve area is not sufficient to contain all the material due to the empirical distribution between rack aisles and product clusters.
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
445
The next section presents several significant improvement activities developed in the Alpha case. Refilling policies along with reserve and picking strategies are interesting areas for improvements in the fulfillment system. Different proposals are developed both in terms of receiving goods and customer order fulfillment. To reduce the risk of LRP, several simulative approaches are set for evaluating and analyzing the reengineering solutions. The logistics reengineering process has proved to be a useful industrial engineering and management technique for achieving significant improvements in operational efficiencies for good quality services in the warehouse/order fulfillment system.
16.5 Reengineering Activities The reengineering process of the picking system starts from the following considerations/assumptions: i. ii.
iii.
iv.
v.
This is a low-level picker to part OPS. In the fast pick area the storage location assignment is fixed, that is, the location for a generic SKU is fixed. This is a single order picking system. The company rejected the idea of organizing the picking and retrieval activities from the fast pick area adopting a batch strategy, e.g. assigning more than one order (or different parts of different orders) to a single picker. The average number of orderlines is very high and does not justify the adoption of new routing procedures and rules in presence of a single order picking strategy: for each customer order the picker visits a large number of locations, i.e. a large number of aisles. The storage capacity of the fast pick area and the storage capacity of the reserve area are constant from the AS-IS to the reengineered system configurations (called TO-BE). The configuration of families/clusters of items and the assignment of families to aisles are the same in AS-IS and TO-BE configurations.
The last two hypotheses can be considered not useful in a reengineering process, but they are necessary because the company did not accept changing the AS-IS configuration of racks which are very variable from one aisle to another. In particular the level of products stored in the forward area significantly changes from one item to the next. The subject of this section is the illustration of the procedure adopted for redesigning the OPS in terms of storage allocation, that is, the determination of the best storage level for each item, and assignment of storage allocation within the fast pick (forward) area and the bulk (reserve) area. The adopted procedure is made up of the following decisional steps as illustrated in Fig. 16.18:
446 Fig. 16.18 Adopted procedure for OPS reengineering
A. Regattieri et al.
• Forward area (AS-IS analysis) capacity • Reserve area (AS-IS analysis) capacity Storage capacity • SKUs or packages' locations mapping on AS-IS and SKUs' locations determination
Storage capacity determination for itemsand families of items
• OPT rule (Bartholdi and Hackmann, 2003)
• rank-based rule for assignment Aisles and locations • SKUs or packages' barycentric coordinates determination assignment in forward area
• distance from SKUs or packages' locations in forward and reserve configurations • number of restocks, • cost of travelling for restocking, KPI measurement • cost of travelling doe picking/retrieving.
1. Determination of the available storage capacity and the percentage of capacity used in the forward and reserve areas. By the previous (iv) assumption the capacity in AS-IS and TO-BE configuration is the same (Storage capacity of an SKU and usage) 2. Mapping of SKUs or packages locations, i.e. determination of the barycentric coordinates, in the AS-IS configuration and for both forward and reserve areas. 3. Determination of the storage capacity, i.e. the maximum admissible level of storage, for a generic item in the forward area adopting a dedicated storage assignment strategy and the single location hypothesis, i.e. at least one location for each item. (Storage allocation of products in the forward area and restocking). 4. Determination of the storage capacity assigned to each family of products; 5. Assignment of aisles and locations to products in the forward area adopting a rank-based rule. (Storage location assignment). 6. Determination of the barycentric coordinates of SKU in TO-BE configuration and for both forward and reserve areas. 7. KPI measurements and comparison (AS-IS vs. TO-BE configurations): distances from packages or SKU locations in forward and reserve configurations; number of restocks; cost of traveling for restocking; cost of traveling for retrieving items. (KPI measurement and comparison (AS-IS vs. TO-BE)). An in-depth analysis of the AS-IS configuration of the system shows that: the historical restock number is very high and depends on the maximum storage level defined for a generic SKU; 99.30% of SKUs, which corresponds to about 7,910 items, have a single fixed location in the forward area. The number of locations for each SKU in the reserve area is very variable, but most items have 1, 2 or 3 different locations. Figure 16.19 illustrates the statistical distribution of the
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
Fig. 16.19 Distribution of the number of locations in the fast pick area
447 120%
2500
100%
Occurence
2000
80% 1500
60% 1000
40% 500
20% 0%
0 1
2
3
5
10
20
30
40
100
50
Altro
Classes
number of locations in the reserve area. This is a typical circumstance in the presence of seasonal products and when a randomized storage assignment is adopted. The historical period of interest is 1 year. The statistical distribution of popularity distinguishes ‘‘very high’’ moving items from ‘‘very slow’’ ones. In particular, in the forward area the number of visits (the total popularity) is 13,078,000 corresponding to 7,966 items moved and 32,731,363 dm3/year (dm3 stays for 1 L). The value of popularity is very variable: from one to more than 5,000 accesses in a year. In the forward (i.e. ‘‘stock’’) area the number of items restocked is 2,760, while the popularity, i.e. the number of visits to the reserve area, is 32,705. Now a brief description of the adopted procedure and the results obtained for some of the steps cited in Fig. 16.18 follows.
16.5.1 Step 1: Storage Capacity of an SKU and Usage The forward (i.e. the ‘‘pick’’) capacity is 3,885,036 dm3 corresponding to 18% of the whole system capacity: 1,237,163 dm3 of this capacity (corresponding to about 32%) is used assuming a ‘‘fluid’’, i.e. continuous, hypothesis of using the storage volume. The reserve capacity is 18,644,922 dm3 corresponding to 82% of the whole system capacity: 5,777,498 dm3 of this capacity (corresponding to about 31%) is used assuming a fluid hypothesis of using the storage volume.
16.5.2 Step 3: Storage Allocation of Products in the Forward Area and Restocking The adopted model for the determination of the fraction of storage volume to be allocated to a single item is the following as proposed by Bartholdi and Hackman (2003):
448
A. Regattieri et al. 120%
1400
1200
100%
1000
Occurence
80%
800 60%
600 40%
400 20%
200
0%
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 10 50 Altro
0
Class of Ratio
Fig. 16.20 Distribution of ratio of allocated capacity in AS-IS and the allocated capacity in TO-BE
pffiffiffiffi f OPTi ¼ P pi ffiffiffiffi i fi With: fi OPTi
ð16:3Þ
rate of material flow through the warehouse for the item i; optimal fraction of available space devoted to item i
Bartholdi and Hackman (2003) demonstrate that this strategy minimizes the number of restocks from the reserve area. Then the minimum number of restocks found is the same for each item i. This could represent a useful guideline for evaluating whether or not the adopted storage allocation strategy is out of balance and produces as many restocks as required. In particular, for this case study, the large variability of restocks for each product in a period of time demonstrates that the OPS is not operating under optimal conditions. For example consider the code 00879006, whose AS-IS capacity is 2,268 dm3 corresponding to 1.48 pallets. The popularity is 2,484 pieces, and the corresponding rate of material flow is about 39,529 dm3/year. Equation (16.3) quantifies a fraction of storage volume equal to 0.069266%, that is about 2,691 dm3 and 1.76 pallets. As a consequence, the allocated capacity for code 00879006 increases by about 18.65% from the AS-IS to TO-BE configuration. Figure 16.20 shows the effect of the reallocation of storage capacity to the whole set of items in the forward area. In particular, the statistical distribution of the ratio of allocated capacity in AS-IS and the allocated capacity in TO-BE is reported. More than 50% of items have an assigned capacity over the optimal
16
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
449
value for restock minimization. The large variability in the rate of materials generates a large variability in the assigned storage capacity. The adopted hypothesis for the reserve area is that the storage capacity of an SKU in TO-BE is the same as in AS-IS. The expected reduction of restocks in one year is about 23.5% corresponding to about 13,396 restocks/year.
16.5.3 Step 5: Storage Location Assignment This section deals with the determination of the storage level assigned to each item in accordance with the assignment of families of products to the aisles, that is, in agreement with the pick list sequence. The pick list is a sequence of retrieval items and orderlines, arranged based on the volume and weight of each bin (see Sect. 16.3.3). For example, during the retrieval process items with high levels of weight and/or low volume (dm3 values) have to be stored at the first levels of picked unit of load. As a consequence most critical items are assigned to the most favorable locations given the pre-assignment of families and items to the aisles (see previous hypothesis (v)). By the reallocation of storage capacity in the forward area the storage level assigned to a generic family of products can significantly change from AS-IS to TO-BE configurations. Given the list of items to be assigned to a specific aisle, a ranking procedure is adopted to define the exact location of each item in the forward area and in accordance with the reallocated storage capacity (see Step 3). The adopted procedure is based on the value of popularity (considering descending values) and not on the value of the so-called index of assignment (AI) defined as numbers of bins moved in a historical period T and adopted by the company in an AS-IS configuration. As a consequence, in a TO-BE system configuration the most visited items are located in accordance with the number of accesses in T, i.e. the number of visits requested of pickers, and not with the flow of bins moved in T (as in AS-IS system configuration). Items with high popularity are located in available and most favorable locations, i.e. near the I/O depot area. According to the aforementioned (v) hypothesis/constraint the features of storage racks and aisles are given and it is possible to modify the assignment of locations to products given the updated values of storage level area and the preassignment of families of items to the most favorable aisles. Following the re-assignment procedure the storage locations refer to a single aisle corresponding to two racks and made of two areas: the first is the low-level area and corresponds to the forward storage volume; the second is the high-level area and corresponds to the reserve area. The basic idea of the adopted greedy rule for storage assignment is to cut the fast pick area and the reserve area of the generic aisle obtaining slides of different widths.
450
A. Regattieri et al.
(a)
(b)
Fig. 16.21 Virtual level and stock and pick areas (a). Slice construction at pick and stock (b)
Family 11 Family 11 Family XX
(a)
(b)
Fig. 16.22 Storage location assignment procedure
Figure 16.21 illustrates this filling process. The warehousing system is made up of several aisles, two racks each, and each aisle has a rack level used for the picked products (see Fig. 16.21a). In a single aisle this level is supposed to be fixed and is called ‘‘virtual level’’. For a given aisle the whole AS-IS area spent for picking is known and is also supposed to be constant in a generic TO-BE. As a consequence the AS-IS configuration defines a virtual level for each aisle, as a couple of racks. The generic rack is cut into slices respecting its virtual level and separating ‘‘picked’’ from ‘‘bulk’’ storage quantities. Different slices represent the assigned storage capacity to different items. The assignment of items is constructed by assigning a single item with another starting from the first available locations close to the I/O depot area (see Fig. 16.21b). Obviously, with two items assigned to the same rack, the corresponding slices have different widths as illustrated in Fig. 16.21b. Literature demonstrates that in the presence of class-based storage allocation of products the right number of classes is three (A, B and C). So in order to reduce the number of classes from the AS-IS configuration made up of 25 different classes, they have been grouped in accordance with the combined values of volume—V, high (H) and low (L), and weight—W, high (H) and low (L). Then the groups of
The Logistics Reengineering Process in a Warehouse/Order Fulfillment System
Fig. 16.23 Storage location assignment and slices configuration
451
STOCK (S)
Virtual level
16
PICK (P)
Y [m]
0
Filling direction Family 11
Barycentric distance (P) & (S)
items of each class of families are assigned to the most suitable aisles in order to pick items with high weight and low volume at the beginning of the picking tour. In fact, all picking missions start in a corner of the warehousing system first visiting aisles that are closer to the depot area. Figure 16.22a and b exemplify the process of assignment families and products to an aisle. The generic family is a tower of bricks, each corresponding to a specific product and made up of two contributions: pick (for the forward storage area) and stock (for the reserve storage area). The assignment is parallel in order to minimize the distance between the picked quantity and the reserve quantity. The priority adopted for the assignment is the popularity: given two or more products to be assigned to an available and favorable location the most ‘‘popular’’ is selected and related slices are identified for both the forward and the reserve areas. The result of the definition of slices at pick and stock is exemplified in Fig. 16.23.
16.5.4 Step 7: KPI Measurement and Comparison (AS-IS vs. TO-BE) Figure 16.24 shows the distribution of distance, in meters, between the pick and stock locations in a specific aisle and for a specific family of products. The mean expected saving on the traveled distance for a single restock is about 6%. But there is another important saving due to the reduction in the number of restocks as illustrated in Sect. 16.5.2. The average cost of a restock is 34.0 m, while the duration, including the variable traveling time and fixed times, is about 8 min and 55 s. The saving distance for a single restock is about 2.2 m. The reduction in restocks in a year generates a saving of 456 km/year. The saving due to the reduction of 1.6 m generates a saving of 95 km/year. The saving time is 2,040 h/year when a picker works about 1,760 h/year. As a consequence the saving corresponds to at least one picker and related annual costs which are about 33 k€/year.
452
A. Regattieri et al.
Occurance
Fig. 16.24 Distribution of distance between pick and stock locations
30
120%
25
100%
20
80%
15
60%
10
40%
5
20%
0
0% 5
10
15
20
25
30
Classes
axis----------------------------------------->
80
--------------------------------------------y
Fig. 16.25 Tour of a picker: AS-IS vs TO-BE. Order #9422 (100 orderlines)
70 60 50 40
TO-BE AS-IS
30 20 10 0
0
10
20
30
40
50
60
