How to Record Warehouse Performance Using DEA

Traditional performance metrics simply don't contain enough information to support comparisons over time or across various sites

DEA enables one warehouse to be compared to a cohort of "peer" warehouses, not based on averages, but based on "best performance
DEA has been implemented in an internet-based tool, iDEAs-W DEA allows warehouse managers to compare their warehouse operations to a large set of other warehouses. iDEAs-W requires only a modest data input effort, and provides a "system efficiency" score

As per DEA scores almost every segment of the warehousing industry, fewer than 20% to 30% of the warehouses are "efficient.

Detailed Analysis
Warehouses use resources (facilities, equipment, inventory investment, labor, etc),to produce an economically valuable service (customer orders shipped) Traditionally, warehouse performance has been measured using a host of single factor performance and single factor productivity metrics Single factor example - lines shipped and fill rate. For example, the output could be lines picked and the input could be labor hours, yielding the labor productivity, or lines per hour The issue : to interpret a particular single factor metric, or to compare two values, one must have a considerable amount of additional information to draw a relation between them

DEA Analysis
DEA would allow a particular warehouse- candidate warehouse--to be compared to a large set of other warehouses. DEA would construct a hypothetical composite warehouse from the input and output data for all other warehouses, and this composite warehouse would be compared to the candidate warehouse The DEA "score" for the candidate warehouse would be reported as a percentage and is interpreted as the composite warehouse used no more than X % of any single resource used by the candidate Warehouse

Limitation
The warehouses compared should be "similar enough" so that they are comparable To work effectively, DEA requires a considerable number of warehouses for comparison;.

Hackman and Frazelle model (1993)

Involved 55 warehouses and the proposed DEA model had the resource inputs and production outputs

The data elements are: Total labor hours includes both direct and indirect labor that is specifically associated with the warehouse function
Lines shipped is simply the total number of lines over all orders Accumulation is defined as the total lines shipped minus the total orders shipped, and is a measure of the extent of sortation required

The Storage Function:

Storage function intended to characterize both the mix of storage types and the space actually needed for Storage

S= a*sqrt(B)+(1-a)*[sqrt(25*P) + sqrt(floor storage sq ft)]

B = number of broken SKU P = Number of Pallet location a = broken case line picked / total line picked

Single Product Factor Metric ( SPFM)

DEA Efficiency Production Frontier

Production/Output

Resource/Input

System efficiency of warehouse B is the ratio OA/OB

Production/Output

Resource/Input

output

Efficient Frontier

Inefficient Frontier

input

Data Envelopment Graph-Frontier Graph

Efficient-frontier with 100% relative efficiency

8

Production/Output

7 6

Best Position

5
4 3 2 1 0 0 0.5 1 1.5 2 2.5 3 3.5

B D

Resource/Input
Note : DEA only gives you relative efficiencies - efficiencies relative to the data considered. It does not, and cannot, give you absolute efficiencies.

Calculation of efficiency of Warehouse D is as below 100(length of line from origin to D /length of line from origin through D to efficient frontier) 100 x ( OD / OP ) = 100 x ( 2.3 / 6.26 ) = 36.7 % Similarly efficiency of B would be 43 %

Warehouse A B C D

Relative Efficiency 100 43 100 36.7

Achieving the efficient frontier reduce its input whilst keeping its output constant (an input target); or increase both its outputs, whilst keeping its input constant (an output target); or do some combination of the above

Partial productivity is the ratio of one output to one input Partial efficiency is the ratio of your partial productivity to the best partial productivity in your peer group.

Lines picked Total labor hours

Your

Lines picked
Total labor hours Lines picked Total labor hours

Best