1. Inventory Optimization-Introduction

2. What is Inventory Optimization Module?
Strategic analysis tool that determines the best cost effective fill rate and optimal stock level for each part/loc
Budget analysis tool to calculate “best” performance given a budget constraint
Enhanced algorithm to fully explore multi-echelon solution space while maintain scalability.

3. The Inventory Optimization Challenge
The Business Challenge:
For more critical or expensive parts, how do we insure that we consider all available information relevant to planning a part within the network on a case-by-case basis, such as:
where to put those parts, and
what levels to invest in to get the maximum “bang for the buck”?
Opportunity:
Use the Servigistics Inventory Optimization to do the additional analysis required to refine initial single item modeling results.

4. The “Big Picture” Inventory Optimization Module
Is intended as a “strategic” solver, used to determine stocking strategy on an infrequent (monthly or less) basis.
To envision the way the Inventory Optimization module works, suppose you have 1000 valuable or critical parts, and 50 stocking locations. That’s 50,000 stocking alternatives. The first stage this module will perform “Marginal Analysis” to pick the part and location that gives the best “bang for the buck” in satisfying customer demand. Then it will conduct “Multi-echelon optimization” to fully explore the solution with the implication of multi-echelon replenishment hierarchy and identify the best solution, The last stage is searching for sub-optimal solution but compliant with budget constraint.
Inventory Optimization expresses its results as ideal ReOrder Point for every part/location pair considered.

5. Part/Loc target is an given input
Item Based Planning
Item Based Planning:
Part/Loc target is an given input
Ignores tradeoff between parts (price)
Ignores tradeoffs between echelons
Ignores “customer wait times”
Net results:
Too much inventory is held for expensive parts
More inventory is held at less crucial location and vice versa

6. System Level Optimization
Field Location
Central Location
Vendor
Field Location
Central Location
Vendor
Inventory Optimized
Inventory Optimized
Planning across multiple levels and multiple parts, providing lowest possible inventory for a target service level
Optimize each part across all locations & all levels, to remove redundancy
Optimize across parts, to limit the investment in slow-moving parts across the chain
Determined by field part’s end-of-support-chain fill rate goal
Based on lowest support chain overall inventory
Results in sub-optimization when used with multiple levels
Optimize a single part for each location
Each stocking location evaluated independently, in sequence
Determined by central & filed fill rate goals
Based on central part’s fill rate goal, demand and LT forecasts, & Safety stocks

7. Inventory Optimization Benefits
Multi-Echelon, Multi-Item planning provides several benefits that most current service parts planning systems are unable to achieve:
For certain types of inventory situations, significant incremental inventory reduction beyond traditional solutions.
Reductions in expediting costs and increases in planner productivity (because higher service levels can be achieved with the same inventory) by price trade off within single echelon, and then echelon trade off when in multi echelons) 7 7

8. Simply put, what happens at the margin
Marginal Analysis
Simply put, what happens at the margin
What are the consequences of spending an additional dollar on spares
For each item at each location calculate Bang4Buck = ΔEBO/$
Benefit divided by price. This is simplified
Choose highest B4B value alternative
Repeat until objective is achieved

9. Use the loss function to determine EBO
Expect Back Order
Use the loss function to determine EBO
If s (inventory position) is 0, then EBO is demand within lead time
The reduce of EBO for adding one unit, is the “benefit” of adding one unit.
Inventory Position is equally distributed between ROP + 1 to ROP + Q (if Q > 1)
Get EBO derivation 9 9

10. Trade Offs Two trade offs need to be understood clearly
Price trade off
Echelon trade off
Trade offs are also the driver of benefits
two sides of the sword, need to be managed.

11. Multi-Item Optimization
97%
96%
92%
99% 0%
Considers the following variables:
Part Cost
Demand
Lead-Time
Demand Variability
Lead-Time Variability
Minimum/Maximum DS
Multi-Item Optimization determines the best tradeoff
between every part’s performance and investment in a location

12. If “benefit” is the same, then cheaper part will yield better b4b.
Price Trade-Off
If “benefit” is the same, then cheaper part will yield better b4b.
If the difference of price for pairs in a scheme is huge, then the trade off will be significant
Say if the most expensive part is 1000 time more than cheapest part. Then expensive part requires 1000 times of benefits before it can be picked before the cheap part
The result of the recommended fill rate/availability could be 1000 times different if all other factors are the same
It is more a business question whether the magnitude of trade off is acceptable.

13. Echelon Trade Off
It should be understood that trade off could increase/decrease either upstream or downstream fill rate/availability dramatically while maintaining the same overall performance
Sometimes, central with low or no demand still could be planned with 99% fill rate while yield overall lower inventory at field locations
Vice versa, due to different lead time and dependent demand pattern, central could only have 50% DS while field locations have 99% will majority of the stock
Even the phenomena could be in smaller magnitude than price trade off, it could still be the major source of saving.

14. Optimal Inventory Solution: Strategic and Tactical
Multi-echelon optimization
Multi-part optimization
Multi-indenture (availability/SLA) optimization
Calculate service level for every part and location
Inventory Optimization
Service Level Targets
Determine ASL for each location
Safety level calculation formulas (type and volume)
Complex part chaining
Pooling
Planning
Stocking Plan (ASL/Levels)
Others, like Subaru of America, have the same situation. With disparate transaction systems and silo planning systems, there is a great opportunity for improvement.
Shortage and Excess thresholds
ASL change management
Balancing
Fair share allocation
Execution /
Management
Actions (buy, repair, move, expedite, cancel, etc.)
Tactical

15. Summary Built on existing Servigistics Platform
Works in combination with Servigistics Planner to provide an optimal strategic and tactical solution
Handles the following real world scenarios:
Multi Echelons of Support
Multi Items
Highly scalable
Applies modeling techniques
Models the entire network
Marginal Analysis (“bang for the buck”)
Incorporates Sherbrook and Muckstadt ‘s Metric Theory
Applies the appropriate statistical distribution to model demand during lead times.
Poisson, Negative Binomial and Normal
Simultaneously considers many MOEs
Type I/II Fill Rate
Budget Constraints - Segment and Pair level override