7/29/2019 Chetan SCM.madras
1/23
SUPPLY CHAIN MANAGEMENT
Submitted By: Group 5Chetan G. Patil / (MS12V016)Piyush Raj / (MS12V017)
Sharad Katwa / (MS12V027)Sitikantha Das / (MS12V028)
Demand Forecasting & Aggregate Planning at GE
Submitted to :
Prof R. P. Sundarraj
7/29/2019 Chetan SCM.madras
2/23
GE BEL
Joint Venture - Nov 1997 GE 74%; BEL 26%
Site Total Area: 350 K Sq. Ft. Built Up Area: 183 K Sq Ft.
700 Employees - 500 GE, 200 Contract
100% Export Oriented Unit
7/29/2019 Chetan SCM.madras
3/23
C-Arm X-ray Radiography Mammography Vascular CTMR
Generator / HV Tanks CT Detector Module
Diverse Application - Catering to wide range of Healthcare products
Products and Applications:
Tubes
7/29/2019 Chetan SCM.madras
4/23
DEMAND :
Problem Statement:
2008 2009 2010 2011
Q1'08 Q2'08 Q3'08 Q4'08 Q1'09 Q2'09 Q3'09 Q4'09 Q1'10 Q2'10 Q3'10 Q4'10 Q1'11 Q2'11 Q3'11 Q4'11
1372 1557 1639 1462 1151 1122 1344 1484 1445 1510 1494 1557 1304 1321 1275 1339
Deman
d
Demand for X Ray tube shows seasonal variations.
Study aims to propose suitable :
(i) Forecasting Model (ii) Aggregate Production Planning Model
1000
1100
1200
1300
1400
1500
1600
1700
Q1'08 Q2'08 Q3'08 Q4'08 Q1'09 Q2'09 Q3'09 Q4'09 Q1'10 Q2'10 Q3'10 Q4'10 Q1'11 Q2'11 Q3'11 Q4'11
7/29/2019 Chetan SCM.madras
5/23
Selection : Holts & Winter Method
Qualitative
Time Series
- Static
- Adaptive
Casual Simulation
FORECASTING
METHOD
Forecasting Methods:
7/29/2019 Chetan SCM.madras
6/23
6
A static method assumes that
the estimates of level, trend
and seasonality within the
systematic component do not
vary as new demand is
observed
In adaptive forecasting, the
estimates of level, trend and
seasonality are updated after each
demand observation.
Time Series Forecasting Methods:
7/29/2019 Chetan SCM.madras
7/23
HOLTs Model:
7/29/2019 Chetan SCM.madras
8/23
8
Regression
Analysis
1000
1100
1200
1300
1400
1500
1600
1700
Q1'08 Q2'08 Q3'08 Q4'08 Q1'09 Q2'09 Q3'09 Q4'09 Q1'10 Q2'10 Q3'10 Q4'10 Q1'11 Q2'11 Q3'11 Q4'11
Estimation of Level & Trend:
7/29/2019 Chetan SCM.madras
9/23
HOLTs Model:
7/29/2019 Chetan SCM.madras
10/23
Here MADt=-1.21. Thus the estimate of standard deviation of
forecast error using holt's model with alpha=.1 and beta=.2 is
-1.515968
9 P9=L8+T8 1664.79
10 P10=L8+2T8 1725.98
11 P11=L8+3T8 1787.18
12 P12=L8+4T8 1848.37
HOLTs Model Forecasted Values:
7/29/2019 Chetan SCM.madras
11/23
WINTER Model :
7/29/2019 Chetan SCM.madras
12/23
Period Demand Deseasonalized Demand SeasonalFactor
1 1151.00 1174 0.980409
2 1122.00 1236 0.907767
3 1344.00 1298 1.035439
4 1484.00 1360 1.091176
5 1445.00 1422 1.016174
6 1510.00 1484 1.01752
7 1494.00 1546 0.966365
8 1557.00 1608 0.968284
Estimation Of Seasonal Factor:
7/29/2019 Chetan SCM.madras
13/23
PeriodT
DemandD1 Level L1 Trend T1
SeasonalFactor
S1Forecas
t F1 Error E1AbsoluteError A1
Mean SquaredError MSE1 MAD % Error MAPE TS1
1112.00 62.00
1.00 1151.00 1174.05 62.01 0.98 1150.52 -0.48 0.48 0.23 0.48 0.04 0.04 -1.00
2.00 1122.00 1237.12 62.22 0.90 1112.45 -9.55 9.55 45.69 5.01 0.85 0.45 -2.00
3.00 1344.00 1299.89 62.33 1.03 1338.32 -5.68 5.68 41.21 5.24 0.42 0.44 -3.00
4.00 1484.00 1362.15 62.32 1.09 1484.83 0.83 0.83 31.08 4.13 0.06 0.34 -3.60
5.00 1445.00 1429.46 63.32 0.98 1396.03 -48.97 48.97 504.48 13.10 3.39 0.95 -4.87
6.00 1510.00 1511.15 66.99 0.90 1344.54 -165.46 165.46 4983.32 38.49 10.96 2.62 -5.96
7.00 1494.00 1565.32 64.43 1.03 1626.11 132.11 132.11 6764.53 51.87 8.84 3.51 -1.87
8.00 1557.00 1609.62 60.40 1.09 1776.33 219.33 219.33 11932.41 72.80 14.09 4.83 1.68
WINTER Model:
7/29/2019 Chetan SCM.madras
14/23
Thus using Winter Model the forecast for next 4
periods is given by the following
P9=(L8+T8)*S1 = 1636.62
P10=(L8+2T8)*S1 = 1557.38
P11=(L8+3T8)*S1 = 1844.55
P12=(L8+4T8)*S1 = 2017.84
Here MADt=72.8. Thus the estimate of standard
deviation of forecast error using winter model
with alpha=.1 and beta=.2 ,gama = .1,is 91
WINTER Model Forecasted Values:
7/29/2019 Chetan SCM.madras
15/23
Forecasting method MAD % Error MAPE
Holt's method -1.213 3.325 4.53
Winter method 72.8 14.09 4.83
Deviation for winter is high
Holts method is the appropriate best model for forecasting.
Analysis : Error Estimates for Forecast
7/29/2019 Chetan SCM.madras
16/23
Aggregate Planning:
Aimed to determine ideal levels of capacity , production ,
subcontracting , inventory , stock outs over a specifiedperiod of time horizon.
Objectives: Determine
ProductionRateSubcontracting
OvertimeWorkforce
Levels
MachineCapacityBacklog
7/29/2019 Chetan SCM.madras
17/23
Demand Forecast Production Costs Labour HRS / unit
Inventory Holding Cost Backlog Cost
Inputs & Constraints:
INPUTS :
CONSTRAINTS :
Overtime Layoffs Capital available
Backlog From Supplier
7/29/2019 Chetan SCM.madras
18/23
Assumptions
October starting Inventory 100 UnitsOctober starting workforce 50 Associates
Working days / month 20
Working HRS / day 8
Cost StructureMaterial cost 50000 per unit
Inventory Holding cost 10000 per unit/week
Cost of stockout/backlog 20000 per unit/week
Hiring & Training cost 5000 per worker
Layoff cost 5000 per worker
Labour HRS required 10 per unit
Regular Time cost 63 perhour
Overtime cost 100 per hour
Cost of subcontracting 1000000 per unit
GE BEL Data:
7/29/2019 Chetan SCM.madras
19/23
Regular time labour cost 10080 x Wt (63/HR x 8 HRS / day x 20
days/month = 10080)
Over time labour cost 100 x Ot
Cost of hiring & lay off - 5000 x Ht + 5000 x Lt
Cost of Inventory & stock out - 10000 It+ 20000 St
Cost of Material & subcontracting - 50000 Pt+ 1000000 Ct
Objective Function :
Linear Programing Model:
10080 x Wt + 100 x Ot + 5000 x Ht + 5000 x Lt+
10000 It + 20000 St + 50000 Pt + 1000000 Ct
Total cost incurred during planning horizon :
7/29/2019 Chetan SCM.madras
20/23
Workforce , hiring & layoff : Wt - Wt-1 Ht + Lt = 0
Capacity constraint : 16 Wt + 0.1 x Ot
Inventory balance constraint : It + Pt + Ct = Dt + St-1 + It + St
Overtime constraint : Ot
7/29/2019 Chetan SCM.madras
21/23
Snapshot:
7/29/2019 Chetan SCM.madras
22/23
Conclusion:
Linear Programing can be used as a flexible
tool, by Operations Manager to meetProduction targets , satisfying all constraints.
7/29/2019 Chetan SCM.madras
23/23