To demonstrate the proposed methodology a small theoretical aircraft supply chain
model, as depicted in Figure 8, is used for analyzing different repair policies. Assume
rate. When a part breaks it is either repaired at the Base or sent upstream to the Depot for
repair.
Assume the questions of interest are 1) How is aircraft availability affected by
increasing the number of parts repaired at the base level? and 2) How is aircraft
availability affected by increasing the number of landing gear assemblies repaired at the
base level?
Figure 8 - Aircraft Supply Chain Example
The first question requires a high level of detail, where the active agents include
Parts, Bases, and Depots. Since Landing Gear and Aircraft agents are used simply to
track availability output there is no process data for these agents, as depicted in Table 9.
Table 9 - Process Parameters for High Resolution Model
Agent Break Rate Repair Rate at Base Repair Rate at Depot Shipment Time to Depot
Part A 15 days 2 days 1 day 2 days
Part B 25 days 5 days 2 days 2 days
Landing Gear -- -- -- --
Aircraft -- -- -- --
For the second question the Parts agents are aggregated to become the Landing
Gear agents. Thus the lower resolution model for the second question includes Landing
agents for this level of resolution. In this simple example the parameters for the Parts
agents were averaged to find the process parameters for the Landing Gear.
Table 10 - Process Parameters for Low Resolution Model
Agent Break Rate Repair Rate at Base Repair Rate at Depot Shipment Time To Depot Part A -- -- -- -- Part B -- -- -- --
Landing Gear 20 days 3.5 days 1.5 days 2 days
Aircraft -- -- -- --
With hierarchical design and object-oriented programming, Aircraft agents form
the super class, with successive subclasses Landing Gear agents, then Parts agents.
Aircraft agents have six fields, or attributes, that correspond to the data specified in the
process parameter tables. Figure 9 provides pseudo code for defining these agents
hierarchically with object-oriented programming.
To demonstrate the use of lookup tables for agent interactions, the same aircraft
supply chain example is used to answer questions regarding repair processes at the Base
level. Assume the questions of interest are now 1) How is aircraft availability affected by
repairing more parts on the flightline versus repairing parts in the backshops? and 2) How
is aircraft availability affected by increasing the number of parts repaired at the Base?
Table 11 shows the agent interactions for repairing the Part agents at the
Flightline and Backshop level, which is the higher resolution model. For the lower
resolution model in the second question, Part agents interact with the Base agents, as
Figure 9 - Aircraft Supply Chain Example Agent Structure
Table 11 - Agent Interactions for High Resolution Model
Agent Repair Message
Sent To
Repaired Message Sent To
Parts (A, B) Flightline / Backshop --
Landing Gear -- --
Aircraft -- --
Base -- --
Flightline -- Parts (A, B)
Backshop -- Parts (A, B)
Table 12 - Agent Interactions for Low Resolution Model
Agent Repair Message
Sent To
Repaired Message Sent To
Parts (A, B) Base --
Landing Gear -- --
Aircraft -- --
Base Depot Parts (A, B)
Flightline -- --
Backshop -- --
Depot EOM Base
As mentioned previously, in the simple case where all parts have the same logic,
gates and switches can be used instead of lookup tables.
4.6 Summary
By combining hierarchical modeling with data-driven modeling the proposed
methodology has extended the variable resolution modeling work to agent based
modeling and simulation (ABMS). Existing literature explains variable resolution
modeling for discrete event simulation, but variable resolution has never been extended
to agent based modeling and simulation (ABMS). This work ties together a general
framework for using ABMS for supply chain risk management, which includes the use of
software agents, for data mining, integrated with agent based simulation platforms. This
framework enables rapid data collection for simulation input, while also providing an
5 Reparable Item supply Chain Risk Measurement Framework
5.1 Overview
The primary difference between consumable and reparable item supply chains is
that reparable items stay in the supply chain until deemed obsolete. Reparable items loop
through the supply chain in a cycle of failures and repairs, which requires more resources
and more extensive management due to the greater complexity. Thus, decision makers
need different information to manage inventory and logistics, and need a whole other
pool of knowledge of repair processes.
To the best of our knowledge, supply chain literature lacks published work in
reparable item supply chain risk metrics. This chapter fills this gap, by providing a risk
metrics framework specific to reparable item supply chains. Along with combining
existing consumable and AF reparable metrics, we developed new metrics and
enhancements to existing metrics. Since the Balanced Scorecard framework has proved
successful in industry and the Department of Defense (DoD), we used its underlying
structure, or categorization, for our reparable item metrics framework.
The AF logistics community uses the balanced scorecard perspective, but has
modified the perspectives from Customer, Internal Business, Finance, and Innovation and
Learning to be Warfighter, Logistics Processes, Resource Planning, and Innovation and
Learning (AFMC 2005). Figure 10 provides the original scorecard designed for
modified scorecard developed under the Expeditionary Logistics for the 21st Century
(eLog21) effort for Department of Defense (DoD) Logistics.
Figure 10 - Original Balanced Scorecard Performance Measures (Kaplan and Norton 1992)
Figure 11 - Balanced Scorecard for DoD Logistics (DoD 2004)
Along with the balanced scorecard, another performance measurement category
from Table 7 that aligns well with the AF is the decision-making levels, i.e. strategic,
tactical and operational. However, these decision making levels are embedded in the
balanced scorecard framework and are apparent when the scorecard is specialized for
different levels of users. The remainder of this chapter discusses consumable item supply
chain metrics, Air Force specific metrics, and our recommended reparable item risk
metrics framework, and aggregation and disaggregation of metrics.