Lot Sizing, MRP Stability, and Planning Performance

Introduction

Material Requirements Planning (MRP) converts a master production schedule into time-phased requirements for components and materials. The explosion logic is conceptually simple, but the resulting planned orders depend strongly on lot-sizing rules, inventory records, lead times, safety stock, and the stability of the planning environment. A technically correct MRP calculation can still produce poor operational results if it creates excessive order changes, unrealistic schedules, or uneconomic order quantities.

Lot sizing determines how much to order or produce when a net requirement occurs. MRP stability concerns whether planned orders remain sufficiently consistent over time for purchasing, production, and suppliers to act efficiently. Planning performance therefore includes both economic measures—such as setup, ordering, holding, and shortage costs—and execution measures—such as schedule changes, expediting, supplier reliability, and capacity feasibility.

This chapter compares major lot-sizing methods and develops a framework for diagnosing nervousness, rescheduling, exceptions, and parameter errors.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Explain the role of lot sizing within MRP.
2. Compare lot-for-lot, fixed-order quantity, period-order quantity, least-unit-cost, least-total-cost, Silver–Meal, and Wagner–Whitin methods.
3. Calculate planned orders using alternative rules.
4. Distinguish economic instability from data and parameter instability.
5. Diagnose nervousness, rescheduling, exception messages, and inappropriate master-data settings.
6. Recommend governance and improvement actions for MRP systems.

1. MRP Lot Sizing as a Decision Problem

For period t, let gross requirements be G_t, scheduled receipts be SR_t, projected available inventory be PAB_t, and safety stock be SS_t. A simplified projected-inventory relationship is:

Here, POR_t is the planned-order receipt in period t. If projected inventory would fall below safety stock, the net requirement is:

The planned-order receipt generated by the lot-sizing rule must cover NR_t, possibly together with requirements in later periods. A planned-order release is offset from its receipt by the item’s lead time. If lead time is L periods, then:

These equations assume deterministic requirements, accurate inventory records, and a valid lead time. In practice, requirements may be probabilistic, receipts may be late, and capacity may prevent the proposed plan from being executed.

Lot sizing balances several costs:

• ordering or setup cost;
• purchase price and quantity discounts;
• inventory holding cost;
• shortage, lateness, or expediting cost;
• transportation and administrative cost;
• capacity and changeover effects.

A method that minimizes inventory cost alone may be inappropriate when setup times, supplier minimums, or finite capacity dominate.

2. Principal Lot-Sizing Methods

2.1 Lot-for-Lot

Lot-for-lot (L4L) sets each planned receipt equal to the net requirement in that period:

It normally produces little or no cycle inventory. Its advantages are simplicity, low average inventory, and close synchronization with changing demand. Its disadvantages are frequent orders, repeated setups, high administrative workload, and vulnerability to demand volatility.

L4L is appropriate where setup costs are low, items are expensive or perishable, requirements are highly variable, or production and replenishment can be changed flexibly. It is often a useful benchmark even when another rule is ultimately selected.

2.2 Fixed-Order Quantity

A fixed-order quantity (FOQ) assigns a predetermined quantity Q whenever a net requirement occurs:

The quantity may be an economic order quantity (EOQ), supplier minimum, container multiple, batch size, or production standard lot. The classical EOQ, under constant demand and simplifying assumptions, is:

where D is annual demand in units per year, S is ordering or setup cost per order, and H is annual holding cost per unit. The formula assumes known and constant demand, instantaneous replenishment, no shortages, constant unit price, and independent replenishment decisions. It is not automatically suitable for time-phased MRP demand.

FOQ can reduce setup frequency and exploit supplier or machine economics. However, it may create excess inventory, especially when demand is intermittent or the selected quantity is much larger than actual requirements.

2.3 Period-Order Quantity

Period-order quantity (POQ) converts an order quantity into a coverage interval. Using an average demand rate d and an order quantity Q, the approximate number of periods covered is:

In practice, T is rounded to a practical number of periods. The order covers net requirements over that interval, beginning with the first uncovered requirement. POQ is more responsive to time-phased demand than FOQ, but it inherits the weaknesses of the estimated average demand and may perform poorly when demand is lumpy.

2.4 Least-Unit-Cost Method

The least-unit-cost method evaluates candidate order spans and selects the span with the lowest average cost per unit. For an order covering periods t through k, let S be the setup cost, h the holding cost per unit per period, and D_j the demand in period j. The candidate cost is:

The unit cost is:

The method adds future periods while unit cost decreases; it stops before a further period increases unit cost. It is intuitive, but the average-cost criterion can favor a large order even when its total cost is not globally optimal.

2.5 Least-Total-Cost Method

Least-total-cost (LTC) compares the setup cost with the holding cost created by extending an order. For a candidate span, the holding cost is:

The method seeks a span where holding cost is approximately equal to setup cost:

It is a balancing heuristic rather than a guaranteed optimum. It is particularly useful when demand is uneven and management wants an understandable compromise between order frequency and inventory.

2.6 Silver–Meal Heuristic

The Silver–Meal method minimizes average cost per period. For an order placed in period t covering through period k, the average cost per period is:

The algorithm begins with k=t, adds one period at a time, and stops when the average cost increases. The previous span is selected. Silver–Meal generally performs well for deterministic, time-varying demand, but it is a heuristic and does not guarantee the global minimum.

2.7 Wagner–Whitin Dynamic Programming

The Wagner–Whitin algorithm finds an optimal replenishment plan for a finite horizon under deterministic dynamic demand. It assumes zero lead time within the planning calculation, no shortages, a fixed setup cost, linear holding cost, and no capacity restriction. Let F(t) be the minimum cost of satisfying demand through period t. Then:

with F(0)=0. The candidate i represents an order in period i covering demand through period t. Backtracking from F(T) identifies the optimal order periods.

Wagner–Whitin is valuable as a benchmark because it explicitly considers the whole horizon rather than making a local decision. Its practical limitations arise when demand is uncertain, capacity is finite, prices vary, lead times are nonzero, or shortages are permitted. Extensions can address some of these conditions, but computational and modeling complexity increases.

3. Worked Comparison

Consider four periods with net requirements of 40, 20, 60, and 30 units. The setup cost is 100 monetary units per order and holding cost is 2 monetary units per unit per period. Assume no beginning inventory, no shortages, and receipts available at the start of each period.

For an order in period 1 covering periods 1–3, holding cost is:

• period 1 demand: 40 units held 0 periods = 0;
• period 2 demand: 20 units held 1 period = 40;
• period 3 demand: 60 units held 2 periods = 240.

Thus, total cost is 100+40+240=380, and total units are 120. Unit cost is 380÷120=3.17. For periods 1–4, an additional 30 units are held three periods, adding 2×30×3=180. Total cost becomes 560 for 150 units, or 3.73 per unit. Therefore, least-unit-cost would stop at period 3.

For Silver–Meal, average cost per period is:

The average cost decreases through period 2 and increases at period 3; Silver–Meal therefore selects an order covering periods 1–2. The contrast illustrates why different heuristics can produce different plans: least-unit-cost uses cost per unit, whereas Silver–Meal uses cost per time period.

A Wagner–Whitin calculation would compare all feasible partitions of the horizon. For example, ordering separately in all four periods costs 4×100=400, excluding holding. Ordering all demand in period 1 costs 100+2(20+2×60+3×30)=480. Other partitions may cost less than either extreme. The dynamic program is preferred when exact deterministic optimization is important.

4. Comparative Evaluation

No method is universally superior. The appropriate choice depends on demand pattern, setup economics, inventory value, supplier constraints, planning horizon, and the cost of instability.

5. MRP Stability and Nervousness

MRP nervousness is excessive change in planned orders caused by small changes in requirements, inventory, lead times, or planning parameters. A plan may be mathematically responsive but operationally unusable. Nervousness can include changing quantities, dates, suppliers, priorities, and cancellation or rescheduling messages.

A useful distinction is between:

• demand-driven change, caused by a genuine customer or production change;
• data-driven change, caused by inaccurate inventory, bills of material, or receipts;
• parameter-driven change, caused by poor lot sizes, safety stocks, or time fences;
• algorithm-driven change, caused by the planning method itself;
• execution-driven change, caused by supplier or shop-floor performance.

Sources of nervousness

1. Forecast updates: requirements near the end of the horizon are repeatedly revised.
2. Lot-sizing discontinuity: a small demand increase can trigger an entire additional FOQ.
3. Lead-time variation: changing lead times shift releases across time fences.
4. Unreliable inventory records: errors appear as artificial shortages or surpluses.
5. BOM and engineering changes: altered component quantities propagate through multiple levels.
6. Late receipts: the system recommends emergency replenishment even when the original order exists.
7. Safety-stock recalculation: frequent policy changes destabilize net requirements.
8. Capacity infeasibility: a theoretically valid plan is repeatedly moved when resources are overloaded.

Stability controls

Important controls include planning time fences, frozen or slushy zones, order-change tolerances, exception thresholds, firm planned orders, supplier schedules, and pegging. A time fence prevents automatic changes within a protected interval unless an authorized planner overrides the rule. A tolerance may ignore immaterial changes, such as a quantity variation below a specified percentage or date variation within an agreed number of days.

Controls must not conceal genuine risk. Excessively wide time fences can preserve a stable but infeasible plan; excessive tolerance can hide shortages and late demand.

6. Rescheduling and Exception Messages

An exception message identifies a condition requiring attention rather than automatically proving that action is necessary. Common messages include:

• expedite an order;
• defer or postpone an order;
• cancel an order;
• increase or decrease quantity;
• release a planned order;
• past-due receipt;
• insufficient lead time;
• material available before requirement date;
• overload or capacity conflict.

The correct response follows a disciplined sequence:

1. Verify the requirement and its source using pegging.
2. Check on-hand balance, allocated inventory, open orders, and transaction timing.
3. Confirm BOM, unit of measure, scrap factor, and lead time.
4. Determine whether the exception is inside a frozen zone.
5. Assess supplier, production, capacity, and customer consequences.
6. Record the decision and its authorization.

Blindly accepting every exception creates instability. Ignoring every exception creates shortages. Effective exception management prioritizes by risk, material criticality, customer impact, and feasibility.

7. Parameter Errors and Diagnostic Logic

Many MRP failures are master-data failures rather than algorithm failures. A diagnostic audit should examine:

A practical test is to select a high-impact item and trace one requirement from the master schedule through BOM explosion, gross requirements, projected balance, netting, lot sizing, and lead-time offset. This “transaction-to-plan” trace often reveals errors more quickly than reviewing aggregate reports.

8. Applied Case: Northstar Pumps (Illustrative)

Background

Northstar Pumps assembles industrial water pumps. A purchased seal is used in every finished pump. Weekly demand is volatile, the supplier ships in cartons of 200, and the supplier’s nominal lead time is two weeks. The MRP system uses FOQ = 200, a safety stock of 100 seals, and automatic rescheduling.

Problem

Planners receive frequent expedite and defer messages. Buyers complain that order dates change several times before release. The warehouse reports high seal inventory, while production occasionally stops because the required seal is not available at the line.

Analysis

The investigation finds four issues. First, receipts are recorded when goods arrive at the central warehouse, but line-side transfers are delayed in the system. Second, the supplier lead time is entered as 10 calendar days although the planning calendar treats it as 10 working days. Third, the 100-unit safety stock is not supported by a documented service-level policy. Fourth, the FOQ of 200 is larger than most weekly net requirements, creating alternating surplus and shortage signals.

A transaction trace confirms that some apparent shortages are caused by location records rather than physical absence. A calendar test shows that releases are systematically late. A demand review shows that the supplier carton multiple is 200, but two cartons can be split operationally, making a fixed quantity of 200 unnecessarily rigid.

Approach and solution

Northstar first corrects inventory-location transactions and changes lead time to two working weeks based on observed supplier performance. It then establishes a one-week frozen zone and a two-week slushy zone. Within the frozen zone, changes require planner authorization. Outside it, changes are accepted only when they exceed defined quantity or date tolerances.

The company retains the supplier’s 200-unit minimum for purchase orders but introduces a supplier-managed delivery schedule that permits staged deliveries. For internal planning, POQ is tested against L4L and Silver–Meal. The team selects a hybrid policy: L4L for requirements inside the frozen zone when stock is available, and Silver–Meal for the unconstrained outer horizon. Safety stock is recalculated from service objectives and lead-time demand rather than an arbitrary round number.

Outcome

After three planning cycles, expedite messages decline, average inventory falls, and the number of order-date changes is reduced. The improvement does not result from selecting a sophisticated algorithm alone. It results from correcting records, aligning calendars, separating firm and flexible planning zones, and matching the lot-sizing rule to supplier and operational realities.

Learning points

1. Stable planning requires trustworthy transaction data.
2. A supplier minimum does not require every internal planning decision to use the same batch logic.
3. Exception messages must be interpreted in operational context.
4. Governance controls are complements to, not substitutes for, accurate parameters.

9. Technology, Analytics, and Governance

Modern enterprise systems can simulate alternative lot-sizing policies, calculate exception priorities, and monitor plan stability. Useful metrics include:

• planned-order change frequency;
• percentage of orders changed inside the frozen zone;
• expedite-message rate;
• schedule adherence;
• inventory turns;
• service level and stockout frequency;
• forecast error and parameter accuracy;
• supplier confirmation and delivery reliability.

AI and advanced analytics can identify unusual parameter combinations, forecast lead-time distributions, classify exception severity, and recommend policy segments. Such tools should support—not replace—planner accountability. An algorithm trained on unstable historical decisions may reproduce poor practices. Governance should require explainability for material decisions, audit trails for overrides, access controls, and periodic validation against service, cost, and feasibility outcomes.

Segmentation is often more valuable than a single enterprise-wide rule. High-value, stable-demand items may use L4L or carefully tuned FOQ; low-value, repetitive items may use larger fixed lots; intermittent items may require special-order policies; constrained components may require capacity-aware planning. The chosen rule should be documented with owner, review frequency, assumptions, and exception conditions.

10. Limitations and Critical Evaluation

Classical lot-sizing methods commonly assume deterministic demand and holding costs. Real systems face uncertainty, substitution, common capacity, supplier minimums, transportation constraints, quantity discounts, and multi-echelon inventory interactions. Holding cost itself may be nonlinear because obsolescence and quality risk rise sharply beyond certain quantities.

Furthermore, minimizing modeled cost can increase total organizational cost if it produces unstable schedules. A more complete objective may be represented conceptually as:

The terms need not be additive in a strict financial sense; the expression emphasizes that planning stability and feasibility should be explicitly evaluated. In research and advanced practice, rolling-horizon optimization, robust optimization, stochastic programming, and simulation can supplement traditional heuristics. Nevertheless, sophisticated optimization cannot compensate for incorrect BOMs, unreliable inventory, or unrealistic lead times.

Best-Practice Framework

A robust implementation follows these principles:

1. Establish clean item, BOM, calendar, and inventory data.
2. Segment items by demand, value, criticality, and supply characteristics.
3. Use L4L as a diagnostic baseline.
4. Compare heuristic results with a full-horizon benchmark where feasible.
5. Protect near-term execution with time fences and tolerances.
6. Review exception messages by material and customer risk.
7. Measure both economic performance and plan stability.
8. Recalibrate parameters using actual performance data.
9. Document overrides and maintain auditability.
10. Treat planners, buyers, suppliers, and production managers as participants in one planning system.

Chapter Summary

Lot sizing determines the quantity and timing pattern of MRP replenishment. Lot-for-lot minimizes cycle inventory but may create frequent orders; FOQ supports batch and supplier economics but can create excess stock; POQ links orders to coverage periods; least-unit-cost, least-total-cost, and Silver–Meal provide understandable heuristics; and Wagner–Whitin gives a finite-horizon optimum under restrictive deterministic assumptions.

MRP nervousness arises when small changes produce repeated changes in quantities or dates. Its causes may be demand volatility, inaccurate records, incorrect parameters, algorithmic discontinuities, or execution failures. Time fences, tolerances, pegging, exception prioritization, and firm planned orders can control instability, but only reliable data and realistic policies provide a durable solution. Planning performance must therefore be evaluated through cost, service, feasibility, and stability together.

Key Terms

Lot-for-lot: A rule that makes each planned receipt equal to the net requirement.

Fixed-order quantity: A rule using a predetermined replenishment quantity.

Period-order quantity: A rule that orders enough to cover a specified number of periods.

Setup cost: Cost incurred when initiating a production run or purchase order.

Holding cost: Cost of carrying inventory over time.

Least-unit-cost: A heuristic minimizing cost per unit over a candidate coverage span.

Least-total-cost: A heuristic balancing setup cost and accumulated holding cost.

Silver–Meal: A heuristic minimizing average cost per period.

Wagner–Whitin: A dynamic-programming method for optimal deterministic finite-horizon lot sizing.

Nervousness: Excessive change in planned orders caused by changing inputs or planning rules.

Time fence: A protected interval limiting automatic plan changes.

Exception message: A system alert identifying a condition requiring review.

Pegging: Tracing a component requirement to its originating demand.

Safety stock: Inventory held to protect against uncertainty or service risk.

Lead time: Time between releasing and receiving an order.

Order multiple: Required increment in order quantity.

Projected available balance: Expected inventory after receipts and requirements.

Net requirement: Requirement remaining after considering available inventory and scheduled receipts.

Master data: Planning records such as BOMs, lead times, calendars, and lot sizes.

Planning horizon: Future interval included in the MRP calculation.

Review Questions

Short Answer Questions

1. What is the principal difference between FOQ and POQ?
2. Why can lot-for-lot create high operating cost?
3. Define MRP nervousness.
4. What information does pegging provide?
5. State two examples of MRP master-data errors.

Long Answer Questions

1. Compare the assumptions, strengths, and weaknesses of Silver–Meal and Wagner–Whitin.
2. Explain how time fences and order-change tolerances influence MRP stability.
3. Discuss why an EOQ may be unsuitable for time-phased MRP demand.
4. Explain how incorrect lead times and inventory records create false exception messages.
5. Develop a governance framework for reviewing lot-sizing parameters.

Analytical / Application Questions

1. A component has requirements of 50, 20, and 80 units in three consecutive periods. Setup cost is 90 and holding cost is 2 per unit per period. Calculate the total cost of one order covering all three periods and compare it with separate orders each period.
2. An item is repeatedly expedited despite apparently adequate warehouse stock. Construct a transaction-trace diagnostic showing at least five records or parameters to verify.
3. A supplier requires a 500-unit minimum, but weekly demand averages 80 units and is highly variable. Recommend a planning and supplier-collaboration policy, comparing FOQ, POQ, and staged delivery options.

Multiple Choice Questions with Answers

Question 1

Which lot-sizing method sets each planned receipt equal to the net requirement in that period?

A. Lot-for-lot

B. Fixed-order quantity

C. Period-order quantity

D. Wagner–Whitin

Correct Answer: A. Lot-for-lot

Question 2

In the classical EOQ formula Q* = √(2DS⁄H), what does S represent?

A. Annual demand

B. Ordering or setup cost per order

C. Annual holding cost per unit

D. Safety stock

Correct Answer: B. Ordering or setup cost per order

Question 3

What is the main criterion used by the Silver–Meal heuristic?

A. Minimum inventory value

B. Minimum cost per unit

C. Minimum average cost per period

D. Maximum service level

Correct Answer: C. Minimum average cost per period

Question 4

Which method uses dynamic programming to find a finite-horizon optimum under deterministic assumptions?

A. POQ

B. Least-unit-cost

C. Silver–Meal

D. Wagner–Whitin

Correct Answer: D. Wagner–Whitin

Question 5

For an order covering periods t through k, demand in a later period creates holding cost because it is received earlier than needed. What determines the number of holding periods for demand in period j?

A. j−t

B. k−j

C. k−t

D. t+j

Correct Answer: A. j−t

Question 6

Which condition is an example of MRP nervousness?

A. A planned order remains unchanged for six cycles

B. A small demand change repeatedly alters order dates and quantities

C. A supplier confirms a purchase order

D. Inventory is physically counted accurately

Correct Answer: B. A small demand change repeatedly alters order dates and quantities

Question 7

What is the primary purpose of pegging?

A. To calculate EOQ

B. To set safety stock

C. To trace a component requirement to its originating demand

D. To determine annual holding cost

Correct Answer: C. To trace a component requirement to its originating demand

Question 8

Which error is most likely to cause systematically late planned-order releases?

A. An incorrect calendar or lead-time setting

B. A correctly recorded receipt

C. A lower-than-expected setup cost

D. A valid order multiple

Correct Answer: A. An incorrect calendar or lead-time setting

Question 9

A one-week frozen zone is intended primarily to:

A. Increase forecast variability

B. Prevent or limit automatic near-term changes

C. Eliminate all safety stock

D. Guarantee unlimited capacity

Correct Answer: B. Prevent or limit automatic near-term changes

Question 10

Which statement best describes Wagner–Whitin?

A. It always handles uncertain demand and finite capacity

B. It is a fixed batch-size rule

C. It is a heuristic based only on average demand

D. It optimizes deterministic finite-horizon replenishment under stated assumptions

Correct Answer: D. It optimizes deterministic finite-horizon replenishment under stated assumptions