Labor Resource Optimization
An important part of a manager's responsibilities is matching the
resource availability with the workload. One of the
"agile" tools I use is the simple chart shown at the left,
drawn for the period of interest, and based on predetermined
categories and on specific individuals' work. Here, 60% of the
workload is represented as devoted to long term
"projects", 15% to relatively short term
"walk-in" jobs, 15% to maintaining what's been generated in
the past, and 10% to continuing education. This breakdown would
mesh with the higher level view, shown at the right, of how an
individual's time is allocated throughout the year (see Typical
Resource Time Allocation). Also in that figure is my breakdown
of incoming jobs into job types, and how those jobs progress through
this job-level view of the Solution Delivery Process. Types are
basically distinguished by resource needs, time frame and number of
interfaces with other Teams/projects. Resources are grouped
according to common job goals, complementary backgrounds and perhaps
according to corporate organizational structure.
All this has been rolled into the high-level resource model shown at
the right, where some of the percentages noted above, as well as the
specific business process, have been changed to meet requirements of
the particular organization being examined. Input parameters
shown include the number of group resources available at simulation
start and the costs of those resources, expected job inter-arrival
rates, and costs incurred when a particular job type has to wait for
resources. Output values (averaged over a one year period)
include the number of jobs entering and leaving, the job span times,
the actual work time spent on each job type (= meeting time + work
time + documentation time), the job wait time, resource utilizations
and the average number of jobs waiting on a resource. The
objective function to be optimized is Cost = ResourceCost +
JobWaitCost.
A part of the validation process for this optimization model is
comparing a range of individual model runs for different resource
levels and costs with the resource optimization results. The
figure at the left shows a range of costs for various numbers of
available resources and a given wait/resource cost ratio. The
cost function is shown in yellow (minimum at ~ 4.3 resources) and the
simulation optimizer predicted optimum number of resources is shown
in black (4.625) -- a difference of ~ 7%. Site managed with MyEclipse, a multi-language, multi-platform IDE