Title: Hybrid Simulation
1Hybrid Simulation Optimization for Navy
Manpower Planning and DistributionDouglas A.
Samuelson, Steven P. Wilcox, Paul Thornton
Navy Manpower Planning ConferenceArlington, VA
6 May 2008
Serco, IncBUPERS-11N00189-04-D-0007-7J50
2Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Introduction
- Background
- ONR-sponsored via Capable Manpower ACQUIRE
program at NPRST - Final phase of Distribution Incentive System
(6.3) research - Purpose
- Model current Enlisted distribution system
business rules and policies - Incorporate alternative paradigms for
distribution and assignment of Enlisted personnel - Optimized assignments
- Demand-driven vice supply-driven requisitions
- Test and compare how variations in policy and
paradigm affect manning, sea-shore rotation, and
other metrics of interest
3Hybrid Simulation Optimization for Navy
Manpower Planning and DistributionOverview
- Task Requirement
- Develop model that baselines the current
distribution system and various implementations
of future systems over an extended period of time - Simulate the current distribution system
- Stimulate proposed system with policy inputs
(user controlled) - Provide capability to vary selected policies
- Produce metrics on Navy manpower utilization
- Integrated data/modeling system
- Computer software product end items
- Software users manual
- Technical Report / Study / Analysis Support
Services
- Task Result/Usage
- Model that
- Baselines the existing distribution system and
policy parameters - Enables analysis of proposed enlisted
distribution policies and requirements across
time - Allows for graphical depiction of manpower
utilization and other manpower metrics
4Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Important
Functions and Capabilities
- Documented current capability
- Project inventory (distributable and
non-distributable) - Develop allocation control to MCA
- Develop manning control (NMP) based on levers
- Develop assignment control
- Developed model to wargame proposed capability
- Set and analyze impact of policy levers on
manpower utilization over time - Generate manpower metrics
- Retain scenarios for comparison
- Objectives
- Fill a higher proportion of high-priority
billets - Reduce the number of misfits
- (placements above or below current skill
level)
5Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Data Testing
- Data Requirements
- Individual enlisted classification and
qualification data (EMF) - Historical distribution outcomes (EMF)
- Force structure information (historical and
future) - History of relevant policy lever settings (such
as accession mix, promotion timing, sea-shore
rotation, MCA prioritization, grade substitution,
etc) - Testing/Verification
- Baseline development tied to current methodology
- 2-spiral development with increase in complexity
and policy variables - Test cases to support integration of features and
modeling components being incorporated - Integrity of implementation of equations
- Verification and Validation
- Behavioral analysis of the model under use
- Joint review and analysis with Navy enlisted
strength planners, - Comparisons of test cases with existing systems
and baseline system
6Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Approaches
Considered
- Markov Chain analytical models
- Get exact calculations
- Some assumptions inappropriate (uniformity of
transition rates) - Can overwhelm computational capacity high
number of states, transitions - Discrete-event simulation
- Straightforward, easily implemented
- Assumes few if any interactions among
individuals, groups - Agent-based simulation
- Handles interactions among individuals, groups,
skill classes - Capable of features difficult to imnplement and
beyond scope of current task - Microsimulation
- Ideal for modeling an aging population age
increases are exact, automatic - Has advantages of Markov Chain analytical models,
more tractable - Not yet used for manpower planning hard to find
good software for this - (best was CoMicSim, TU-Koblenz not documented
in English yet)
7Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Approach
Adopted
- Discrete-event simulation with agent-based
capability - Java-based, interfacing to MASON (ABS package
from GMU) - Assumes no interactions among individuals, groups
but could handle - Could readily be adapted to full microsimulation
- Simulation projects available manpower by skill
set, for specified years - User can specify up to 15 years, whatever EMCs
are desired - Projects available, attrition, promotions using
pre-specified rates - No interaction between simulation and
optimization - Optimization handles assignments for each year
- Typical assignment problem with prioritization of
infeasibilities - User can adjust specification of objectives /
priorities and rerun as desired
Easy to use and interpret Faithful to actual
situation Easy to maintain, upgrade and extend
8Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Initial Screen
9Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Add New
Scenario
10Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Edit Scenario
11Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Specify
Policies for Filling Positions
12Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Specify
Policies Detail (Pay Grade Substitution)
13Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Specify
Policies Additional Detail (Pay Grades)
14Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Specify
Assignment Policy
15Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution View Billets
16Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution View Inventory
17Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution View
Inventory Detail
18Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Configure
Rates
19Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Reports Menu
20Hybrid Simulation Optimization for Navy
Manpower Planning and Distribution Conclusions
- SimOpt provides
- A framework to expand the rules within the
requisition and assignment modules - Ability to add rules and enhancements to the
simulation of the inventory - Reports and metrics describing the results from
the simulation - Recommended next steps
- See how well the model tracks actual events
- Note suggestions for improved utility, ease of
use - Expand and improve the requisition process
- Improve ultimate assignments
- Take group effects, interactions among EMCs into
account - Incorporate interactions between assignments and
future availability - Incorporate responsiveness to changing external
conditions - The authors gratefully acknowledge the support
and guidance of the Office of Naval Research
(ONR) and Navy Personnel Research, Studies and
Technology (NPRST).