Inventory Management

Material Requirements Planning

Chris Caplice
ESD.260/15.770/1.260 Logistics Systems
Oct 2006
 Assumptions: Basic MRP Model
Demand Discounts
„ Constant vs Variable „ None
„ Known vs Random „ All Units or Incremental
Continuous vs Discrete
„
Excess Demand
Lead time „ None
„ Instantaneous „ All orders are backordered
„ Constant or Variable „ Lost orders
(deterministic/stochastic)
„ Substitution
Dependence of items
Perishability
„ Independent
„ None
„ Correlated
„ Uniform with time
„ Indentured
Planning Horizon
Review Time
„ Single Period
„ Continuous
„ Finite Period
„ Periodic
„ Infinite
Number of Echelons
Number of Items
„ One
„ One
„ Multi (>1)
„ Many
Capacity / Resources
Form of Product
„ Unlimited
„ Single Stage
„ Limited (Constrained)
„ Multi-Stage
MIT Center for Transportation & Logistics – ESD.260 2 © Chris Caplice, MIT
 How many components are there?

Image of iPod Shuffle circuitry removed due to

copyright restrictions.

MIT Center for Transportation & Logistics – ESD.260 3 © Chris Caplice, MIT
 Traditional Management

Purchasing Production Marketing

MIT Center for Transportation & Logistics – ESD.260 4 © Chris Caplice, MIT
 Supply Chain Integration

Information / Planning

Materials Physical
Management Distribution

Vendor Purchasing Production Marketing Customer

MRP MRP MRP MPS DRP DRP DRP

Inventory Deployment

Material Requirements Planning

Master Production Scheduling
Distribution Requirements Planning
MIT Center for Transportation & Logistics – ESD.260 5 © Chris Caplice, MIT
 Inventory Management so far . . .
Traditional techniques . . .
„ Forecast demand independently for each item based
on usage history
„ Establish lot sizes independently for each item based
on demand forecasts
„ Establish safety stocks independently for each item
based on forecast errors
Which make the following assumptions . . .
„ Demand is "Continuous“
[usage occurs in every period]
„ Demand is "Uniform"
[average usage per period is stable over time]
„ Demand is "Random"
[usage in any given period is not known in advance]

MIT Center for Transportation & Logistics – ESD.260 6 © Chris Caplice, MIT
 Cycle Stock with a Fixed Lot Size
A= $500, r=25%, v= $50,
D = 2000 units/yr, Q*=400 units
600
600
On Hand Inventory

400
400

Demand
200
200

0
0

Problem: Intermittent Demand

4 production periods, 500 units/period,
Demand rate 2000/year
MIT Center for Transportation & Logistics – ESD.260 7 © Chris Caplice, MIT
 Fixed Lot Size with Intermittent
Demand results in . . .
On Hand Inventory 600

400

200

Can we do better?
MIT Center for Transportation & Logistics – ESD.260 8 © Chris Caplice, MIT
 Another Wrinkle . . . Product Indenture
Bicycle
Model 1234

Frame Front Wheel Saddle Rear Wheel Handlebars

Fork Rim Shaft Rim Bar

Front fender Axle Seat Axle Gooseneck
Rear Fender Spoke Cover Spoke Grip
Sprocket Tire Tire
Crank Tube Tube
Pedal Sprocket
Chain guard

Note that each item, sub-assembly, component

etc. might feed into multiple end products
MIT Center for Transportation & Logistics – ESD.260 9 © Chris Caplice, MIT
 Combined Demand Impacts
Suppose a widget is part of three items
„ Product A – 10 items per week – (3 Weeks OH)
„ Product B – 5 items per week – (2 Weeks OH)
„ Product C – 7 items per week - (4 Weeks OH)
End demand looks like . . .

1 2 3 4 5
A 10 10 10 10 10
B 5 5 5 5 5
C 7 7 7 7 7
Widget 22 22 22 22 22
MIT Center for Transportation & Logistics – ESD.260 10 © Chris Caplice, MIT
 Combined Demand Impacts
But if ordered separately – what will widget
demand look like?
„ Product A – 10 items per week – (3 Weeks OH)
„ Product B – 5 items per week – (2 Weeks OH)
„ Product C – 7 items per week - (4 Weeks OH)
1 2 3 4 5
A 30 0 0 30 0
B 10 0 10 0 10
C 28 0 0 0 28
Widget 68 0 10 30 38

Important to synchronize
MIT Center for Transportation & Logistics – ESD.260 11 © Chris Caplice, MIT
 Push versus Pull Systems
Simple Example
„ You make shovels that have 4 parts:
Š Metal Digger
Š Wooden Pole
Š 2 Screws
„ Production is 100 shovels per week:
Š Metal part is made in 400 item batches on first 2 days of the month
Š Handles are procured from Pole Co.
Š Assembly occurs during first week of each month
„ How should I manage my inventory for screws?
Š A=$0.25, v=$0.01, r=25%
Š D = 800*12=9600 units per year
Š L = 1 week
„ What are the values for . . .
Š Q* =
Š xL =
Š RMSE(L) =

MIT Center for Transportation & Logistics – ESD.260 12 © Chris Caplice, MIT
 Push versus Pull Systems
What is my policy if I follow a . . .
„ Standard EOQ policy?
Š Order ~1385 (~every other month)
Š What would the Inventory On Hand look like?
„ Standard (s,Q) policy?
Š So, since σL = 193, pick a CSL=95% k=1.64
Š s=185 + (1.64)193 = 502 units
Š Order 1385 units when inventory position ≤ 502
„ Standard (R,S) policy?
Š Select a monthly review policy (R=4 weeks)
Š xL+R= 9600/(52/5) = 923 units
Š σL+R = 193(√5) = 432 units
Š S = 923 + (1.64)432 = 1631
Š Order up to 1631 units every 4 weeks
„ Other methods?
MIT Center for Transportation & Logistics – ESD.260 13 © Chris Caplice, MIT
 Material Requirements Planning
Push vs Pull Systems
„ Push – MRP
Š “initiates production in anticipation of future demand”
„ Pull – JIT
Š “initiates production as a reaction to present demand”

Major Premises
„ Inventory control in a production environment
„ Many products, many component parts
„ Complex product indenture structure
„ Production creates "lumpy" demand

Major Concepts
„ Dependent demand versus independent demand
„ Requirements calculation versus demand forecasting
„ Schedule flow versus stockpile assets
„ Information replaces inventory

MIT Center for Transportation & Logistics – ESD.260 14 © Chris Caplice, MIT
 Material Requirements Planning
Primary Questions
„ What are we going to make? => use forecast
„ What does it take to make it? => use res. req’s & BOM
„ What do we have? => use inventory records
„ What do we need and when? => use mfg schedules
Information Requirements
„ Master Production Schedule
„ Product Indenture Structure
„ Inventory Status
„ Ordering Data
MRP Process
„ Requirements Explosion
„ Use of Bill of Materials (BOM)
„ Net from Gross Requirements
„ Requirements Time Phasing
„ Planned Order Release
MIT Center for Transportation & Logistics – ESD.260 15 © Chris Caplice, MIT
 Example: Bike Co.
BOM Explosion

Bicycle Level 0

Wheel (2) Crank Assembly (1) Level 1

Spoke (86) Sprocket (1)

Tire (1) Crank (2) Level 2
Pedal (2)

MIT Center for Transportation & Logistics – ESD.260 16 © Chris Caplice, MIT
 Bill of Materials Weekly buckets

Product Sub-assembly Component Quantity Lead Time

Bicycle [1] 2
Wheel 2 1

Spoke 86 3
Tire 1 2

Crank Asm 1 1
Sprocket 1 4
Crank 2 3
Pedal 2 3

MRP Approach:
1. Start with Level i demand (i=0)
2. Find Gross Requirements (GR) and On Hand (OH) for Level i
3. Find Net Requirements (NR) for Level i+1 (NR=GR-OH)
4. Establish Planned Order Release (POR) for Level i using Level i lead times
5. Set GR for Level i+1 based on POR for Level i
6. Set i = i+1 and go to Step 2
MIT Center for Transportation & Logistics – ESD.260 17 © Chris Caplice, MIT
 The MRP Plan for the Bicycle
Objective:
Have materials ready for having 25 bikes in week 8
ITEM PERIOD: 1 2 3 4 5 6 7 8
Gross Bicycle Rqmt 25
Requirement On Hand
Due In 25
POR 25
On Hand
>> >> >> >> >> >> >> >>
Wheel Rqmt 50
Net On Hand
Requirement Due In 50
POR 50
>
Spoke Rqmt 4300
Planned
On Hand
Order Release
Due In 4300
POR 4300

MIT Center for Transportation & Logistics – ESD.260 18 © Chris Caplice, MIT
 Item Period: 1 2 3 4 5 6 7 8
Bicycle GR 25
OH
NR 25
POR 25
W heel GR 50
OH

Level 0 NR
POR 50
50

Spoke GR 4300
OH
NR 4300
POR 4300
Tire GR 50

Level 1 OH
NR 50
POR 50
Crank Asm GR 25
OH
NR 25
POR 25
Sprocket GR 25

Level 2
OH
NR 25
POR 25
Crank GR 50
OH
NR 50
POR 50
Pedal GR 50

What is missing?
OH 20 20 20 20 20
NR 30
POR 30

MIT Center for Transportation & Logistics – ESD.260 19 Ordering Plan © Chris Caplice, MIT
 Two Issues
How do we handle capacity constraints?

How do we handle uncertainty?

MIT Center for Transportation & Logistics – ESD.260 20 © Chris Caplice, MIT
 Approach: Optimization (MILP)
Decision Variables: Data:
Qi = Quantity purchased in period i Di = Demand per period, i = 1,,n
Zi = Buy variable = 1 if Qi>0, =0 o.w. Co = Ordering Cost
Bi = Beginning inventory for period I Chp = Cost to Hold, $/unit/period
Ei = Ending inventory for period I M = a very large number….

MILP Model
Objective Function:
• Minimize total relevant costs
Subject To:
• Beginning inventory for period 1 = 0
• Beginning and ending inventories must match
• Conservation of inventory within each period
• Nonnegativity for Q, B, E
• Binary for Z

MIT Center for Transportation & Logistics – ESD.260 21 © Chris Caplice, MIT
 Approach: Optimization (MILP)
Objective Function
n n
Min TC = ∑ CO Z i + ∑ CHP Ei
i =1 i =1

s.t.
Beginning & Ending
B1 = 0 Inventory Constraints
Bi − Ei −1 = 0 ∀i = 2,3,...n
Conservation of
Ei − Bi − Qi = − Di ∀i = 1, 2,...n
Inventory Constraints
MZ i − Qi ≥ 0 ∀i = 1, 2,...n
Ensures buys occur
Bi ≥ 0 ∀i = 1, 2,...n only if Q>0
Ei ≥ 0 ∀i = 1, 2,...n
Non-Negativity &
Qi ≥ 0 ∀i = 1, 2,...n Binary Constraints
Z i = {0,1} ∀i = 1, 2,...n

MIT Center for Transportation & Logistics – ESD.260 22 © Chris Caplice, MIT
MRP: Example
MRP: Example
MRP: Example
MRP: Example

-
I Tighten blnalng
MRP: Example
I
-
Notes:
I
Solves the End Items and the Components models separateltc
What is the impact? insight?
who n 7
 Handling Uncertainty
Safety Stock
„ Add to existing stock levels
„ Where would this be applied?

Safety Times
„ Pad the planned lead times
„ Where would this be applied?

MIT Center for Transportation & Logistics – ESD.260 28 © Chris Caplice, MIT
 Optimal Lead Time Padding
Let:
t = Delivery Time, a random variable
t' = Forecasted Delivery Time
σ = Standard Deviation of the Forecast Error
Tp = Padded Lead time = t' + kσ
Q = Lot Size in units
v = Unit Cost
r = Holding Cost per unit per time period
Cd = Shortage Cost per time period
Tp ∞
TC[T p ] = ∑rvQ( T p - t)P[t ] + ∑ C (t - T
d p )P[t ]
t=0 t=T p+1

* Cd
CSL =
C d + rvQ
MIT Center for Transportation & Logistics – ESD.260 29 © Chris Caplice, MIT
 Optimal Lead Time Padding

⎛ Tp ⎞ ⎛ Tp ⎞ ⎛ ∞ ⎞ ⎛ ∞ ⎞
TC[T p ] = ⎜⎜ ∑rvQT pP[t ] ⎟⎟ − ⎜⎜ ∑rvQtP[t ] ⎟⎟ + ⎜ ∑ C dtP[t ] ⎟ − ⎜ ∑ C d T pP[t ] ⎟
⎜ ⎟ ⎜ t=T +1 ⎟
⎝ t=0 ⎠ ⎝ t=0 ⎠ ⎝ t=T p+1 ⎠ ⎝ p ⎠

dTC[T p ] ⎛ Tp ⎞ ⎛ ∞ ⎞
= rvQ ⎜⎜ ∑P[t ] ⎟⎟ − ( 0 ) + ( 0 ) − ⎜ C d ∑ T pP[t ] ⎟ = 0
dTp ⎜ t=T +1 ⎟
⎝ t=0 ⎠ ⎝ p ⎠

rvQ ( Prob[ NoStockout ]) − ( C d ( Prob[ Stockout ]) ) = 0

rvQ ( CSL *) = Cd (1 − CSL *)

*Cd
CSL =
C d + rvQ
MIT Center for Transportation & Logistics – ESD.260 30 © Chris Caplice, MIT
 Optimal Lead Time Padding
Example:

v = $5.00/unit Q = 1000 units

r = 36% annual t' = 10 days
rv = .005 dollars/unit/day σ = 3 days
Cd = $500 per day (t ~ normal)

CSL* =

k* =

Tp* =

MIT Center for Transportation & Logistics – ESD.260 31 © Chris Caplice, MIT
 Benefits of MRP
Lower Inventory Levels
„ Able to better manage components
„ Increased visibility
Fewer Stock outs
„ Relationships are defined and explicit
„ Allows for coordination with MPS
Less Expediting
„ Due to increased visibility
Fewer Production Disruptions
„ Input needs are explicitly modeled
„ Plans are integrated

MIT Center for Transportation & Logistics – ESD.260 32 © Chris Caplice, MIT
 Shortcomings of MRP
MRP is a scheduling, not a stockage, algorithm
„ Replaces the forecasting mechanism
„ Considers indentured structures
MRP does not address how to determine lot size
„ Does not explicitly consider costs
„ Wide use of Lot for Lot in practice
MRP systems do not inherently deal with uncertainty
„ User must enter these values – by item by production level
„ Typical use of "safety time“ rather than "safety stock“
MRP assumes constant, known leadtimes
„ By component and part and production level
„ But lead time is often a function of order size and other activity
MRP does not provide incentives for improvement
„ Requires tremendous amount of data and effort to set up
„ Initial values are typically inflated to avoid start up issues
„ Little incentive to correct a system “that works”
MIT Center for Transportation & Logistics – ESD.260 33 © Chris Caplice, MIT
 MRP: Evolution of Concepts
Simple MRP
„ Focus on "order launching“
„ Used within production – not believed outside
Closed Loop MRP
„ Focus on production scheduling
„ Interacts with the MPS to create feasible plans
MRP II [Manufacturing Resource Planning]
„ Focus on integrated financial planning
„ Treats the MPS as a decision variable
„ Capacity is considered (Capacity Resource Planning)
Enterprise Resource Planning Systems
„ Common, centralized data for all areas
„ Implementation is costly and effort intensive
„ Forces business rules on companies

MIT Center for Transportation & Logistics – ESD.260 34 © Chris Caplice, MIT
 Questions?
Comments?
Suggestions?