1. Measurement of Instability In Planning Systems
K. Erkan KABAK
Department of Industrial Engineering
Dokuz Eylul University

2. OUTLINE INTRODUCTION INSTABILITY MEASURES NERVOUSNESS
MAJOR CAUSES OF NERVOUSNESS
UNDESIRABLE EFFECTS OF NERVOUSNESS
STABILITY, FLEXIBILITY, ROBUSTNESS
THE NEED FOR A RESEARCH ON NERVOUSNESS
MEASUREMENT OF INSTABILITY
INSTABILITY MEASURES
IN MRP ENVIRONMENT
A NUMERICAL EXAMPLE TO MEASURE INSTABILITY
IN SUPPLY CHAIN ENVIRONMENT

3. OUTLINE (Cont’d) PROPERTIES OF AN IDEAL METRIC MEASURING INSTABILITY
TYPES OF CHANGES LIKELY TO OCCUR IN AN MRP ENVIRONMENT
A NEW METRIC FOR MEASURING INSTABILITY IN MULTI-LEVEL MRP SYSTEMS
FACTORS AFFECTING INSTABILITY
STRATEGIES TO DAMPEN THE MRP NERVOUSNESS
CONCLUDING REMARKS
REFERENCES

4. INTRODUCTION Nervousness
The MRP systems could have some problems with nervoussness,which is often defined as,
Significant changes in production plans, which occur even with minor changes in higher-level MRP records (or MPS), (Vollmann et al. 2004),
However, in literature different terms such as schedule instability or stability terms are used interchangeably to imply the nervousness in MRP plans.

5. INTRODUCTION Major Causes of Nervousness
Major causes of nervousness in schedules are:
Uncertainties about supply and/or demand,
Dynamic lot-sizing,
Rolling planning.
(Steele 1975, Whybark and Williams 1976, Mather 1985,
Kadipasaoglu and Sridharan 1997)

6. INTRODUCTION A snapshot of causes of instability(
Rejected Material 3%
Supplier Allocation Changes 11%
Gain & Losses 22%
Material Transfers 5%
Over/Under Production 1%
Over/Under Shipments 19%
Scrap 19%
Service Requirements 1%
Unscheduled disbursements 1%
Engineering Changes 19%
Figure 1: A snapshot of causes of instability (Inman and Gonsalvez,1997).

7. INTRODUCTION Undesirable effects of Nervousness (incoming)
Some of the undesirable effects of nervousness in the company are:
Increased production and inventory costs,
Increases in throughput times
Reduced productivity,
Decrease in capacity utilization,
Lower customer service levels,
Confusion on the shopfloor,
Loss of confidence to planning system,
Low morale, etc.

8. INTRODUCTION Undesirable effects of Nervousness (outgoing)
From the supply chain point of view, instability forces suppliers to react to unexpectedly changing requirements, thus increasing cost through increased:
overtime,
undertime,
Inventory,
premium freight,
changeovers,
material handling,
record keeping,
disruptions in quality,
throughput improvement efforts and,
managerial intervention (Inman and Gonsalvez,1997).

9. INTRODUCTION Stability
A stable schedule is one that does not often change with time as additional requirements data are added to the planning horizon (Sridharan et al. 1988),
Actually, stability implies a given production plan will be followed, whereas instability means not following the plan,
Besides stability, there are other stability-related concepts to evaluate the sensivity of planning methods e.g., flexibility and robustness.

10. INTRODUCTION Flexibility Robustness
Generally, the ability of a system to respond to unexpected situations is defined as “flexibility” and it comes out as an important competitiveness factor for companies (Heisig,2002).
Robustness
Heisig (2002) defines robustness as the invariability of initial decisions, or the persistence of complete decision sequence. However, from the planning point of view, it is understood as the insensitivity of the planning systems when the parameters change in stochastic input data.

11. INTRODUCTION
Stability is a matter of balancing flexibility and robustness
If the MPS is stable, stable production plans are obtained, and in consequence stable production,
If too few changes are made, this will result in bad customer relations and increased inventory becaused of poor flexibility,
On the contrary, if too many changes are made, this will result in a reduced productivity,

12. INTRODUCTION Measurement of instability
Measurement of instability would enable performance comparison of different planning procedures for managing the production plans under a variety of operating conditions,
In constructing a measure for schedule stability, it is easier to measure schedule instability. The frequency and magnitude of changes in orders over planning horizon could provide a measure of schedule instability.

13. INTRODUCTION Need for a Research on Nervousness
Maintaining a stable production plan is an important competitive factor in material requirements planning system,
The American Production and Inventory Control Society (APICS) has listed “improved MRP” systems as one of the top 10 topics of concern to its over 80,000 members in 2000.

14. INTRODUCTION Need for a Research on Nervousness
The size of the market for MRP-based production planning and control software is over US $ billions,
With regard to the supply chain management (SCM), Heisig (2002) states that one of the most important enablers for efficient supply chain operations is schedule stability. Moreover, in a Just-in Time (JIT) environment, it causes significant problems for suppliers,
These facts justify the need for additional research for improving the performance of MRP systems,

15. Measurement of Instability
“Before stability can be improved, it must be measured” (Inman and Gonsalvez, 1997)
“You can not control what you do not measure” (Mather, 1994)
“As with any management endeavour, a prequisite to success is measurement”
(Johnson and Davis,1995)

16. INSTABILITY MEASURES In MRP environment:
BKM metric (Blackburn, Kropp, Millen, 1986),
SBU metric (Sridharan, Berry, Udayabhanu, 1988),
Unahabhokha, Schuster, Allen, (2002),
KS metric (Kadipasaoglu and Sridharan, 1996),
Kimms (1998),
De Kok and Inderfurth (1997),
Heisig (2002),
Pujawan (2004),

17. INSTABILITY MEASURES In supply chain environment:
Inmann and Gonsalvez (1997),
Van Donselaar, K., Van den Nieuwenhof, J., and Visschers, J., (2000) and,
Meixell (2003)

18. BKM Metric It is proposed by Blackburn, Kropp, ve Millen (1986),
Instability is defined as the number of times an unplanned order is made in the first period when the schedule is rolled forward,
Shortcomings of BKM metric:
It fails to assess changes in open orders within the cumulative lead time. Thus, considering only the immediate period schedule changes is not adequate,
It depends on enumeration and as the planning horizon length increases, it quickly approaches to the desired value of zero, i.e., as the horizon length is increased, instability artificially is reduced,
It is assumed that demand is deterministic and lead times are zero, therefore the need for open order rescheduling is questionable.

19. SBU Metric It is proposed by Sridharan, Berry ve Udayabhanu (1988),
Instability is defined as the weighted average of schedule changes in order quantity per order over subsequent planning cycles ,
Differences between SBU and BKM metric:
SBU metric applies a weighting procedure to schedule changes, whereas BKM metric does not,
Decreasing weights are used to represent the increased ability to respond increasing uncertainty of demands,
Weights can be varied, for example a small value for the weighting parameter α can be placed for rapidly decreasing weights, whereas a larger value for α can be placed for nearly equal weights so that the instability measure could be adjusted to reflect the importance of the immediate future.

20. SBU Metric t = time period,
= scheduled order quantity for period t during planning cycle k,
= beginning of planning cycle k,
N = planning horizon length,
S = total number of orders over all planning cycles, and
= a weight parameter to represent the criticality of changes in schedule; 0<α<1

21. SBU Metric
Furthermore, they state the instability measure in terms of the proportion of an average MPS order that is changed (A). As approximating for this average they use the well known Economic Order Quantity (EOQ) formula. The modified instability measure is then given as
This standardized measure is based on the assumption that the average order cycle, or natural cycle (T), is equal to the ratio of the EOQ and average or expected period requirements (R), respectively, i.e.

22. SBU Metric Shortcomings of SBU metric
The single-level product structure, i.e., this measure of instability is intended for single-item, single-level situations, but the components at lower levels must also be considered. However, Zhao and Lee (1993) used this metric for multi-items,
It is biased. Because this metric divides total instability by the total number of orders, measures instability as “weighted instability per order”. Also, number of orders depends on the cost structure of the item, and on the ratio between the setup cost and the holding cost (S/h),

23. SBU Metric Shortcomings of SBU metric
The instability would be less if the S/h ratio is smaller and the it would be greater if the S/h ratio is larger. Because when the ratio S/h is small the total number of orders is greater, and the less instability,
The instability measure is not normalized between values of 0 and 1, i.e., maximum stability and maximum instability.

24. Unahabhokha et al. (2002)
Unahabhokha, Schuster and Allen (2002), use a version of the SBU metric presented by Sridharan et al. (1988),
In this measure, weight function is changed by weighting factor and instability is measured for multi items,
i = item number,i= 1,….,n
t = time period, weeks,
= scheduled production for item i, period t during
planning cycle 2,
H = planning horizon length, H=1…52,
k = multiple factor (=100),
Wt = weighting factor, Exp(1/t)-1,
B = total production for planning cycle 2,

25. KS Metric
Kadipasaoglu and Sridharan (1996) have extended the previous instability measure SBU metric and eliminated some shortcomings of it by adding a weight parameter β to assign decreasing weights to the levels,
j = item level, j = 0, ..., m (based on low level coding),
i = item i at level j , i = 1, ...nj,
= the order quantity (open and/or planned) for item i at level j in period t during planning cycle k,
=weight parameter for all levels; 0< β <1.

26. KS Metric
This metric defines instability as the weighted change in order quantities for all items at all levels through the subsequent planning cycles,
Shortcomings of KS metric:
It is still not standardized between a minimum and maximum value of nervousness,
In addition, different kinds of change are not still differentiated. For example, the weights for the setup and quantity changes are the same.

27. Kimms (1998)
He considers a T-period problem on the MPS level which is rolled n>0 times,
As different from other measures, he adds a frozen zone, ΔT. The plan for the first ΔT periods is implemented and a new plan is then generated for the periods ΔT+1,…, ΔT+T which coins the name rolling horizon,
For two subsequent planning cycles i and i-1, instability is the weighted difference between the production plans for periods t+(i-1)ΔT and t+iΔT,

28. Kimms (1998)
Instability for an item j, can be defined as the maximum instability:
or the mean instability:

29. Kimms (1998) i = number of run,
= weighted production quantities for item
j, temporarily scheduled , i =1,…,n-1,
= weighted production quantities for item
j, after rescheduling in overlapping
periods, i =2,…,n-1,
= production quantity for item j in period t,
= item-specific weights for item j in period t ,

30. Kimms (1998)
Some possibilities of the instability measure for the overall production plan:
the maximum of the maximum instabilities:
the mean of the maximum instabilities:
the maximum of the mean instabilities:
the mean of the mean instabilities:

31. Kimms (1998)
Relations these four performance measures are given in the following two inequalities:
(1)
(2)

32. Kimms (1998) Shortcomings:
Kimms gives a different weight parameter, it is item independent but this weight can not be varied as Sridharan’s parameter α does,
Only the end item level is considered, but the components at lower levels also must be considered since their schedules can significantly impact both material and capacity plans,
From the standardization point, although the differences in quantities are divided by a maximum in the equation, Kimms’ measure can take on values greater than 1.

33. De Kok and Inderfurth (1997)
They separate the short-term planning stability with long-term stability,
Furthermore, stability examined on the aspect of pure qualitative changes or quantitative changes of order decisions,

34. De Kok and Inderfurth (1997)
Setup stability;
Quantity stability;
= the expected value,
= stationary inventory control rule,
= cumulative demand function with
a stochastic iid demand D.
For
For

35. De Kok and Inderfurth (1997)
It is concluded that the reorder point s does not influence stability whereas the lot size determining parameters Q, S-s, and T can have a considerable impact.
Shortcomings:
The stability measures are restricted to planned order deviations only referring to the most imminent period in each planning cycle, i.e., they describe short-term stability,
Stability measures are not for multi items and do not take into account the impact of the product structure.

36. Heisig (2002)
Heisig (2002), proposes the modified setup and quantity instabilities,
Setup oriented instability:
= replenishment order size in period t as planned in
cycle j,
N = number of planning cycles,and
α = a weight parameter to represent the criticality of plan
revisions; 0<α<1

37. Heisig (2002) Quantity oriented instability:
= maximum quantity that can be changed between two
successive planning cycles,
= maximum reasonable demand per period,

38. Pujawan (2004)
Pujawan (2004) presents a case study of schedule instability based on a field observations and a model to quantify the instability,
In his model, there are three types of changes that could happen in the day-to-day production schedules:
(1) Changes in the production start time,
(2) Changes in the specification/design of items
being ordered and,
(3) Changes in the quantity committed to be
delivered.

39. Pujawan (2004)
(1)
(2)
i = type of changes (i.e., 1: production start time;
2: specification/design; 3: quantity),
j = period in the planning horizon,
t = planning cycle,
k = order,
wi = weight of type i change,
= the quantity of order k which in the previous cycle was scheduled
to be produced in period j but then experiencing type i change when
observed in planning cycle t,
= aggregated instability observed in planning cycle t from all orders
where in the previous planning cycle they were scheduled in period j,
= total instability observed in planning cycle t,

40. Pujawan (2004)
In determining the weights, the managers were asked to do pair-wise comparison for the three types of changes. The results then converted into weights using the logic of analytical hierarchy process (AHP). The final weights are obtained as follows: w1= 0.10, w2= 0.62 and w3= 0.28,
When a change is involved in a more than one type, the one with the larger weight is considered.

41. Pujawan (2004)
Figure 2: Initial production schedule
(for week 39)
Figure 3: Updated production schedule
(for week 40)
From figures 2 and 3, aggregated instability observed in planning cycle of week for all orders previously scheduled in weeks 40 to 43 as follows:
I(40,40) = 0.28*200 =56, I(40,41) = 0.1*430 =43,
I(40,42) = *500 =50, I(40,43) = 0
I(40) = =149

42. A Numerical Example To Measure Instability In MRP Environment
To test the metrics, an example provided by Kadipasaoglu et al. (1997), is considered,
The demand forecast is generated from normal distribution with a mean of 100 units and standard deviation of 20 units.
Table 1: Item Master and Bill of Material (BOM) data. A 2C B 1
LFL 3 C 2 A B 4
POQ
Q-P
Comp.
Parent LS
LSR
LLC SS ST LT
P-NO
Bill Of Material
Item Master
Figure 4: Product
structure
Table 2: On-hand and Scheduled Receipts data for the first planning cycle.
608 54 C
304 27 B
341
163 A 12 11 10 9 8 7 6 5 4 3 2 1 OH
P-NO

43. Table 3(a): MRP records in the beginning of the first and second planning cycles
(actual demand in period 1 is 31 units more than anticipated).
Planning cycle 1
Planning cycle 2

44. Table 3(b): MRP records in the beginning of the second and third planning cycles
(actual demand in period 2 is 79 units more than anticipated).
Planning cycle 2
Planning cycle 3

45. Table 3(c): MRP records in the beginning of the third and fourth planning cycles
(actual demand in period 3 is 51 units less than anticipated).
Planning cycle 3
Planning cycle 4

46. A Numerical Example to Measure Instability
Table 4(a): Planned and open orders
for item A in all cycles.
Table 4(b): Summary of orders
for item A in all cycles. 2
Number of planned
orders in the last
cycle 1
orders released in
the first period of
the cycle 3
orders***
Number of open orders postponed
Number of open
orders expedited
Number of open orders scheduled in the first cycle
orders** 4
Number of
cancelled orders 5
Number of new orders
Number of total
orders*

47. A Numerical Example to Measure Instability
Table 5(a): Planned and open orders
for item B in all cycles.
Table 5(b): Summary of orders
for item B in all cycles. 1
Number of planned
orders in the last
cycle 2
orders released in
the first period of
the cycle 3
orders
Number of open orders postponed
Number of open
orders expedited
Number of open orders scheduled in the first cycle
Number of
cancelled orders 4
Number of new orders 6
Number of total

48. A Numerical Example to Measure Instability
Table 6(a): Planned and open orders
for item C in all cycles.
Table 6(b): Summary of orders
for item C in all cycles. 1
Number of planned
orders in the last
cycle 2
orders released in
the first period of
the cycle 3
orders
Number of open orders postponed
Number of open
orders expedited
Number of open orders scheduled in the first cycle
Number of
cancelled orders 4
Number of new orders 6
Number of total

49. A Numerical Example to Measure Instability
Table 7: Summary of total orders for
all items in four subsequent planning cycles 4
Number of planned orders in the last cycle 5
Number of planned
orders released in
the first period of
the cycle 9
Number of total
released planned orders 3
Number of open orders postponed
Number of open
orders expedited
Number of total open orders scheduled in the first cycle 8
Open orders 10
Number of
cancelled orders 13
Number of total new orders 17
Orders
Figure 5: Percentage of orders with changes observed in three planning cycles.

50. Measurement of instability with the BKM metric
In regard to BKM metric, instability is measured by counting the number of times new orders are scheduled in an imminent period of the planning cycle,
In the example, the BKM value for instability would be three, due to an unplanned order for item A in period 2 during planning cycle 2 and the expedited orders for items B and C in period 3 during planning cycle 3.

51. Measurement of instability with the SBU metric
Table 8: Measurement of instability with the SBU metric.

52. Measurement of instability with the KS metric
Table 9: Measurement of instability with the KS metric.

53. INSTABILITY MEASURES In supply chain environment:
Inmann and Gonsalvez (1997),
Van Donselaar, K., Van den Nieuwenhof, J., and Visschers, J., (2000) and,
Meixell (2003)

54. Inman and Gonsalvez (1997)
They state that from the supplier’s point of view the most harmful type of demand variability is schedule instability,
They measure both the incoming and outgoing schedule instability for a first-tier automotive supplier,
They emphasize that reducing instability is especially advantageous for vertically integrated manufacturers,

55. Inman and Gonsalvez (1997)
Instability is measured by a percentage deviation for each part,
HI_REQ = the highest requirement for the part,
LO_REQ = the lowest requirement for the part,
INIT_FORECAST = the initial forecast for the part,
Horizon = how far into future wanted to look,
Base_week = week contains the most recent timely schedule,

56. Table 10: Example of Material Planning Schedule
Inman and Gonsalvez (1997)
Table 10: Example of Material Planning Schedule … n m
7/22/96 (Week 1) l k j
7/15/96 (Week 2) i h g f
7/08/96 (Week 3) e d c b a
7/01/96 (Week 4)
3 Weeks
From Now
2 Weeks
Next
Week
This
FORECASTS FOR
ACTUAL
REQUIRED
WEEK OF
RELEASE
Example 1: If Horizon = 4 and Base_week = 0,
HI_REQ = max{e,i,l,n,m};
LO_REQ = min{e,i,l,n,m}; INIT_FORECAST = e,
Example 2: If Horizon = 4 and Base_week = 1,
HI_REQ = max{e,i,l,n};
LO_REQ = min{e,i,l,n}; INIT_FORECAST = e,

57. Van Donselaar et al. (2000)
They state that supply chains can become very nervous, if the planning of them are based on dynamic lot sizes and the MRP logic,
To measure the instability, only the setup oriented instability is focussed. In other words, only the number of setups per period is measured,
In their study, it is concluded that three variables have a strong impact on this potential nervousness of MRP:
the lot-sizes,
the demand uncertainty and,
the product structure,
They found that in the worst case (i.e. small lot sizes, low demand uncertainty and a divergent product structure) the MRP planning was 17 times more nervous than its alternative, called LRP (Line Requirements Planning),
On average MRP was 4 or 5 times more nervous than LRP.

58. Meixell (2003)
Meixell (2003) observes the impact of capacity, commonality and batching on schedule instability,
Findings of her study suggest that there is an intensifying effect of order batching, a stabilizing effect of component commonality, and an augmenting effect of limited capacity on schedule instability.

59. Meixell (2003)
Scheduled instability is measured as the coefficient of variation across the scheduled quantities, X, across multiple releases, m, and for a specific production period t. The measure simply states:
This measure may be applied to the first period production only, to each schedule period in the horizon independently, or to some weighted average across all periods.

60. Figure 6: 8-Week Rolling Schedule Example
Meixell (2003)
For example, one production week-number 5 is presented in the multiple release example in Figure 5, the stability for the schedule quantity series CV(47, 37, 56, 20) =
Figure 6: 8-Week Rolling Schedule Example

61. Properties of an ideal metric measuring instability
(1) The measure should be normalized between zero and one, i.e. maximum stability and maximum instability,
(2) The new metric should depend on instability sources in order to measure instability more objectively.
In an MRP environment, there are two sources of instability in the schedules (Kadipasaoglu et al., 1997). The first source is demand uncertainty and the other is rolling planning,
Their effects are shown in Figure 7.

62. Properties of an ideal metric measuring instability
DEMAND
UNCERTAINTY
ROLLING
PLANNING
OPEN ORDER
SCHEDULE CHANGES
AT END ITEM LEVEL
PLANNED ORDER
SCHEDULE
CHANGES
AT LOWER LEVELS
SOURCES OF INSTABILITY
Figure 7: Sources of instability and their effects.

63. Properties of an ideal metric measuring instability
(3) As distance from the beginning period of cycle increases, uncertainty of demands in later periods of planning horizon and ability to react to such demand increase. The weights in the metric should reflect this ability,
(4) The ideal metric should take the relationship between the end item and component levels into account,
(5) Finally, the measure is, as far as possible, independent of the chosen parameters of the planning framework (Heisig, 2002).

64. Types of changes likely to occur in an MRP environment
The types of changes likely to occur in MRP environment are classified into 3 groups:
(1) Pure time changes,
(2) Time and quantity changes,
(3) Pure quantity changes.
These types of changes are given in Table 11.

65. Table 11: Types of changes likely to occur in an MRP environment
No quantity change
Quantity increase
Quantity decrease
ODD=0 and NDD=0
C. Pure Quantity Changes
Advanced (expedited) order
If NDD<ODD
Postponed order
If NDD>ODD
ODD<NDD or ODD>NDD
B. Time and Quantity Changes
Cancelled setup
If ODD>0 and NDD=0
New setup
If ODD=0 and NDD>0
ODD=0 or NDD=0
A. Pure Time Changes
Explanation
Type of change
ODD = Original due date, NDD = New due date,
= order quantity in i period t planning cycle

66. A NEW METRIC FOR MEASURING INSTABILITY IN MULTI-LEVEL MRP SYSTEMS
(1)
(2)
for
= stability,
= total instability,
= quantity instability for planned orders,
= quantity instability for scheduled receipts,
= quantity instability for expedited released planned
orders,
= setup instability for scheduled receipts,

67. A NEW METRIC FOR MEASURING INSTABILITY IN MULTI-LEVEL MRP SYSTEMS
(3)
(4)

68. A NEW METRIC FOR MEASURING INSTABILITY IN MULTI-LEVEL MRP SYSTEMS
(5)
(6)

69. A NEW METRIC FOR MEASURING INSTABILITY IN MULTI-LEVEL MRP SYSTEMS
Period weight function for planned order
Period weight function for scheduled receipts
Weight function for levels
Weight function for expedited and postponed orders

70. FACTORS AFFECTING INSTABILITY
Operating conditions (environmental factors):
Demand uncertainty,
Forecast error,
Capacity tightness,
Safety stock method,
Item cost structure (or natural cycle),
Product structure,
Lot sizing method,
Rolling horizon parameters:
Planning horizon,
Replanning periodicity,
Freezing proportion

71. No-capacity constraint Deterministic/Stochastic
Table 12: Summary of findings on the factors affecting instability
Authors (years)
Settings
Results
Sridharan, Berry,
and Udayabhanu (1988)
Deterministic
Single-item
No-capacity constraint
Freezing proportion
Planning horizon length
Natural cycle (TBO)
Instability
Sridharan
and Berry (1990)
Stochastic
Replanning periodicity
Forecast error
and Lafarge (1994)
Inventory
Service level
Zhao et al. (2001), Xie, Zhao ve Lee (2003)
Deterministic/Stochastic
Multi-item
Capacity constraint
The selection of lot sizing rule has a significant impact on schedule instability
Kadipasaoglu and Sridharan (1997)
Cost structure (TBO) Forecast error Product structure
Unahabhokha, Schuster, Allen (1987)
Forecast bias Safety stock Capacity Forecast error and
safety stock

72. Strategies To Dampen Instability In MRP Systems
Blackburn et al. (1986) proposed and investigated five different strategies for treating instability caused by the interaction of lot sizing decisions and the planning horizon under deterministic demand:
(1) Freezing the schedule within the planning horizon,
(2) Lot for Lot after stage 1,
(3) Safety stocks,
(4) Forecast beyond planning horizon,
(5) Change cost procedure
The results indicate that safety stock and lot-for-lot strategies are not cost effective and when measured solely in terms of instability, freezing the schedule within the planning horizon appears to be dominant.

73. Strategies To Dampen Instability In MRP Systems
Vollmann et al (2004) introduces three guidelines reducing instability in MRP plans:
(1) Freezing and time fences,
(2) Selective use of lot-sizing procedures,
(3) Using firm planned orders,

74. Table 13: Overview of methods for reducing instablity, Heisig (2002)
Orientation
Methods
Lot-sizing
“stable” lot sizing rules
Inventory
Oriented
Buffering
Safety stock
Safety lead time
Safety capacity
Overplanning
Eliminating
causes of
Nervousness
Rolling horizon schedule parameters, e.g. freeze MPS
Forecasting beyond planning horizon
Control engineering changes
Eliminate transaction errors
Minimize supply uncertainty
Local-oriented
Demand management
Time-fencing
Lead-time compression
Pegged requirements
Firm planned orders
Dampening
procedures
Static dampening procedure
Automatic rescheduling procedure
Cost-based dampening procedure

75. Freezing Method
Vollmann et al (2004) states that in MRP systems, the approach of freezing the MPS, especially over the cumulative lead time of a product, is frequently used in practice,
The freeze length interval, called ΔP, is the proportion of the overall planning horizon that is frozen,
A variant of this procedure is the freezing of all decisions within the planning horizon, i.e. P = ΔP. Then, there are no plan revisions (Heisig, 2002),

76. Freezing Parameters
The planning horizon (N): is defined as the number of periods for which the production schedules are developed in each replanning cycle. The planning horizon is expressed as a multiple (K) of the natural ordering cycle (T),
The frozen interval is the number of scheduled periods within the PH for which the schedules are implemented according to the original plan. The free interval is the number of scheduled periods beyond the frozen interval.
The replanning periodicity (RP): is the number of periods between successive replannings.
The greater the RP, the less frequently replanning will occur and computational requirements will be reduced. However, a higher RP may also increase both the probability of capacity-related unfeasibility(more loss of sales) and total cost,

77. Freezing Parameters
The freezing method: two methods exist in freezing the MPS; a period based method(P1, freeze P periods and roll 1 period) and an order based method (JJ, freeze J orders and roll J orders) (Sridharan, Berry, and Udayabhanu 1987, 1988),
The freezing proportion (F): refers to the ratio of the frozen interval relative to the planning horizon (N). It equals to P/N under period based freezing method, whereas it equals to J/K under order based freezing method.
The higher the freezing proportion, the more stable the production schedule, and the lower the schedule instability of the MPS system. However, a higher FP may also increase the loss of sales (under capacity constraint) and the total cost.

78. Figure 8: Demonstration of MPS Freezing Parameters .

79. Table 14: Summary of findings on the impact of freezing parameters
Authors (years)
Settings
Planning Horizon (PH)
Freezing
Proportion
(F)
Replanning Periodicity
(RP)
Sridharan, Berry,
and Udayabhanu (1987)
Deterministic
Single-item
No-capacity constraint TC -
and Udayabhanu (1988) SI
Sridharan
and Berry (1990)
Stochastic
and Lafarge (1994)
Inventory SL
Zhao et al. (2001)
Multi-item
Capacity constraint
Xie, Zhao ve Lee (2003)

80. Concluding Remarks
In this presentation, stability and related concepts, i.e., flexibility, robustness, nervousness, are explained. The need for measuring the instability of planning systems is identified,
By measuring instability, performances of different production plans could be easily compared and evaluated,
It is observed that none of the metrics reflects all the properties of instability and each of them have some shortcomings,

81. Concluding Remarks
Measurement of the instability is represented by an detailed example proposed by Kadıpasaoglu and Sridharan (1997),
Properties of an ideal metric is defined and a new metric to measure instability in multi-item, multi level MRP environment is proposed,
There are several factors affecting schedule instability. These are environmental and rolling horizon factors. The summary of findings of them are given in Table 12,
Furthermore, several alternative strategies which have been suggested in literature to dampen the instability in MRP systems are presented in Table 13,
One of the most frequently encountered approaches is freezing a portion of the schedule. The summary of findings of them are given in Table 14.

82. References
Blackburn, J. D., Kropp, D. H., Millen, R. A., Comparisons of strategies to dampen nervousness in MRP systems. Management Science, 32, , 1986.
De Kok, T., and Inderfurth, K., Nervousness in inventory management: Comparison of basic control rules. European Journal of Operational Research, 103, 55-82, 1997.
Heisig, G. Planning stability in materials requirements system. Springer-Verlag Berlin Heidelberg, 2002.
Inman, R.R., and Gonsalvez, D.J.A., Causes of Schedule Instability in an Automotive Supply Chain. Production and Inventory Management, 37, 2nd Quarter, 26-31, 1997.
Jensen, T., Measuring and improving the planning stability of reorder- point lot-sizing policies. International Journal of Production Economics 30-31, , 1993.
Kadipasaoglu, S. N., and Sridharan, V., Measuring of instability in multi-level MRP systems. Int. J. Prod. Res., Vol. 35, No. 3, , 1997
Kimms, A., Stability Measures for Rolling Schedules with Applications to Capacity Expansion Planning, Master Production Scheduling, and Lot Sizing. Omega, Int. J. Mgmt. Sci. Vol. 26, No.3, pp 355 – 366, 1998.

83. References
Mather, H., Dynamic lot-sizing for MRP: help or hindrance? Journal of Production and Inventory Management, 2, 29-35, 1985.
Meixell, Mary, J., The impact of Capacity, Commonality, and Batching on Schedule Instability: An Exploratory Study. Working Paper, 2003.
Pujawan, N., Schedule nervousness in a manufacturing system: a case study. Production Planning and Control, Vol. 15, No: 5, , 2004.
Silver, E. A., Pyke D. F., and Peterson R., Inventory Management and Production Planning and Scheduling, 3rd Edition, John Wiley and Sons, 1998.
Sridharan, V., Berry, W. L., Udayabhanu, V., Freezing the Master Production Schedule under rolling planning horizons. Management Science, 33 (9), , 1987.
Sridharan, V., W.L. Berry and V. Udayabhanu., Measuring Master Production Schedule Stability under Rolling Planning Horizons. Decision Sciences 19, , 1988.
Sridharan, V., Berry, W. L., Freezing the master production schedule under demand uncertainty. Decision Sciences, 21 (1), , 1990.
Sridharan, V., LaForge, R. L., Freezing the master production schedule: Implications for Fill Rate. Decision Sciences, 25 (3), , 1994.

84. References
Steele, D. C., The nervous MRP system: how to do battle. Journal of Production and Inventory Management, 16, 83-89, 1975.
Unahabhokha, C., Schuster, E. W., Allen, S. J., Master Production Schedule Stability under Conditions of Finite Capacity, 2002.
Van Donselaar, K., Van den Nieuwenhof, J., and Visschers, J., The impact of material coordination concepts on planning stability in supply chains. International Journal of Production Economics, 68, , 2000.
Vollmann, T. E., Berry, W. L., Whybark, D. C., Manufacturing planning and control systems. 4th edition, Irwin, Homewood, III, 1997.
Xie, J., Zhao, X., Lee, T. S., Freezing the master production schedule under single resource constraint and demand uncertainty. International Journal of Production Economics, 83, 64-84, 2003.
Whybark, D. C., and Williams, J. G., Material Requirements Planning Under Uncertainty. Decision Sciences, , 1976.
Zhao, X., Lam, K., Lot-sizing rules and freezing the master production schedule in material requirements planning systems. International Journal of Production Economics, 53, , 1997.
Zhao, X., Xie, J., Jiang, Q., Lot-sizing rule and freezing the master production schedule under capacity constraint and deterministic demand. Production and Operations Management, 10, no. 1, 45-67, 2001.

85. THANKS FOR LISTENING