Replenishment Systems and Policy Design

Introduction

Inventory replenishment is the set of decisions and procedures used to restore stock after consumption. A replenishment policy specifies when an order is triggered, how much is ordered, and what information and controls govern the decision. The policy must balance competing objectives: high product availability, low investment, limited ordering and transport cost, protection against uncertainty, and operational simplicity.

For a postgraduate materials manager, replenishment is not merely an application of an economic order quantity. It is a policy-design problem involving demand characteristics, supplier performance, lead-time risk, service commitments, capacity, information quality, obsolescence and governance. A mathematically attractive policy can perform poorly when its assumptions do not match the supply environment.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Distinguish continuous-review and periodic-review replenishment systems.
2. Design and compare min–max, base-stock and two-bin policies.
3. Calculate order quantities, reorder points, order-up-to levels and review intervals.
4. Incorporate demand and lead-time uncertainty into safety-stock decisions.
5. Select policies using cost, service, risk and operational criteria.
6. Diagnose implementation failures and design appropriate controls.

1. Replenishment Policy as a Control System

A replenishment system has four elements:

• State measurement: on-hand stock, inventory on order, backorders and demand history.
• Decision rule: the mathematical or procedural rule that determines an order.
• Response: supplier or internal production replenishment.
• Feedback: receipts, stockouts, forecast revisions and parameter review.

The relevant inventory position is usually more useful than physical stock alone:

where IP is inventory position (units), OH is on-hand inventory, OO is inventory on order, and BO is backorders or confirmed shortage demand. The equation assumes that all three quantities are measured at the same time and in the same unit. In systems where allocated stock is not included in on-hand inventory, committed allocations must also be subtracted or otherwise represented.

A policy must also define the review event. In a continuous-review system, stock status is monitored after each transaction or frequently enough to approximate continuous observation. In a periodic-review system, stock is assessed only at fixed intervals. The choice affects responsiveness, information-system requirements, order consolidation and safety stock.

Important terminology

• Lead time: elapsed time between placing an order and its usable receipt.
• Cycle stock: inventory expected to be consumed between replenishments.
• Safety stock: inventory held to absorb uncertainty in demand, lead time or supply.
• Reorder point: inventory-position threshold that triggers replenishment.
• Order-up-to level: target inventory position to which an order raises stock.
• Service level: a specified measure of supply performance; its definition must be explicit.
• Stockout: inability to satisfy demand immediately from available stock.
• Backorder: demand recorded for later fulfilment rather than lost immediately.

2. Demand, Lead Time and Service Definitions

Policy design begins with a demand and supply diagnosis. Demand may be stable, seasonal, intermittent, promotional, dependent on a production schedule, or affected by a product life cycle. Lead time may be deterministic, variable, capacity-constrained or affected by customs and transportation risks.

Two service measures are particularly important:

1. Cycle service level (CSL): probability of no stockout during a replenishment cycle.
2. Fill rate: proportion of total demand supplied immediately from stock.

These measures are not interchangeable. A policy can avoid a stockout in most cycles but still have a poor fill rate if occasional shortages are large. Management should specify whether the relevant objective concerns orders, demand units, production continuity or customer response time.

For approximately normal independent demand with mean demand per period μ_d and standard deviation σ_d, demand during a fixed lead time L has:

These relationships assume independent, identically distributed period demand and a fixed lead time. If demand is autocorrelated, seasonal, intermittent or non-normal, direct use of these equations may be inappropriate.

If both demand and lead time vary, a common approximation for the variance of lead-time demand is:

where E[L] is expected lead time and Var(L) is lead-time variance. This form assumes demand periods are independent and that demand and lead time are independent. If they are correlated, a covariance term is required and the approximation should be replaced by simulation or a more appropriate stochastic model.

For a target cycle service level represented by standard-normal quantile z, safety stock is commonly calculated as:

and the reorder point is:

Here μ_DL and σ_DL are mean and standard deviation of demand during the protection period. The protection period is normally lead time in continuous review and review interval plus lead time in periodic review. The formula assumes a normal approximation and that a CSL objective is appropriate. It does not directly guarantee a particular fill rate.

3. Continuous-Review Systems

A continuous-review policy observes inventory position continuously and orders when a threshold is reached. The familiar [? formula is an [? (s,Q) policy:

• s is the reorder point.
• Q is the order quantity.
• When IP ≤ s, place an order for Q units.

In practice, the system may use a transaction trigger rather than literally continuous observation. The policy is most suitable for important, expensive or critical items for which timely information and rapid response justify monitoring cost.

3.1 Order quantity

For relatively stable independent demand, the basic economic order quantity is:

where D is annual demand (units/year), S is ordering or setup cost per order (currency/order), and H is annual holding cost per unit (currency/unit/year). Unit consistency gives units for Q^*. The model assumes constant demand, instantaneous replenishment, no shortages, constant lead time, and a linear holding-cost approximation. It is a benchmark, not a universal answer.

If unit purchase price is constant and no quantity discount applies, purchase cost does not affect the EOQ decision. If discounts, transport minimums, supplier capacity, shelf life or production batch sizes exist, the feasible order quantity must be evaluated against those constraints.

The approximate annual relevant cost is:

This includes ordering and cycle-stock holding costs but excludes purchase cost and safety-stock cost unless those are added separately. At the unconstrained EOQ, annual ordering cost approximately equals annual cycle-stock holding cost.

3.2 Reorder point

For deterministic demand and lead time:

where d is demand per time unit and L is lead time in the same time unit. With uncertainty, the ROP includes safety stock as shown earlier. The order quantity and reorder point solve different problems: Q controls replenishment size and order frequency, while ROP controls protection against lead-time demand.

Worked Example 1: Continuous review

An engineering component has annual demand of 12,000 units. Ordering cost is £45 per order, and annual holding cost is £3 per unit. Daily demand is approximately 40 units, lead time is 5 working days, and the desired CSL is 95%. Daily demand standard deviation is 12 units. Assume 250 working days per year and fixed lead time.

First, calculate the order quantity:

Q^* = √(2 × 12,000 × 45 / 3) = √360,000 = 600 units.

Mean lead-time demand is μ_L = 40 × 5 = 200 units. Lead-time standard deviation is σ_L = √5 × 12 ≈ 26.83 units. For 95% CSL, z ≈ 1.645, so:

SS = 1.645 × 26.83 ≈ 44.15 units.

Rounding upward for protection, safety stock is 45 units and:

ROP = 200 + 45 = 245 units.

The policy is therefore approximately an (s,Q) policy of (245, 600): order 600 units when inventory position reaches 245 units. The calculations assume demand is sufficiently close to normal during lead time and that shortages are not intentionally backordered.

4. Periodic-Review Systems

A periodic-review, or fixed-interval, system reviews inventory every R time units. At each review, the order raises inventory position to an order-up-to level S. If inventory position immediately before review is IP, the order is:

The protection period is R + L, because no new order is placed until the next review and the order then requires lead time to arrive. For variable demand and fixed review and lead time:

This formula assumes independent demand with constant mean and standard deviation per period, fixed lead time, and a normal approximation. More generally:

The periodic system generally requires more safety stock than a continuous system because it must cover demand during the review interval as well as lead time.

Choosing the review interval

A review interval may be imposed by supplier schedules, route frequency, production planning or administrative practice. Where a cost-based approximation is suitable, an interval can be related to an order quantity:

where Q is units per replenishment and D_t is demand per unit of time. This is a conversion, not an independent optimization rule. Review cost, supplier minimums, vehicle utilization, risk and workload must also be considered.

Periodic review is useful when many items are ordered from the same supplier, when route-based purchasing is economical, or when continuous transaction monitoring is impractical. Its weaknesses include delayed response and greater exposure to demand surges between reviews.

5. Min–Max and Order-Up-to Policies

A min–max policy defines a lower limit min and an upper limit max. When inventory position reaches or falls below the minimum, the order raises inventory position toward the maximum:

The precise trigger may be continuous or periodic. Thus, “min–max” is a policy family rather than a single algorithm. In a continuous implementation, min often resembles a reorder point and max resembles reorder point plus an order quantity. In a periodic implementation, the minimum may represent protection-period demand and the maximum an order-up-to level.

Min–max policies are intuitive and effective for large item populations. However, poor parameter governance can create excessive stock: a high maximum combined with a low trigger may produce large average inventories, while a low maximum may cause repeated emergency orders.

A useful parameter relationship for a continuous min–max design is:

This is a design convention, not a law. It is appropriate when an order of approximately Q is intended after the trigger and when inventory position is used consistently.

6. Base-Stock Policies

A base-stock policy attempts to maintain inventory position at a target level B. After each demand or replenishment event, the system orders enough to restore the position:

If orders cannot be negative, the operational form is q_t = max(0, B − IP_t). Base-stock systems are especially suitable for high-value or critical items, short replenishment cycles and environments where frequent small orders are feasible.

For continuous review with fixed lead time, a base-stock level is often set near expected lead-time demand plus safety stock:

Unlike an (s,Q) system, a base-stock system does not deliberately create a large cycle-stock batch. It can therefore reduce response delay but may increase ordering, transaction and transport costs. In multi-echelon systems, the base-stock level must be assigned to the correct echelon; simply adding local targets can duplicate protection stock.

7. Two-Bin Systems

A two-bin system divides stock into two physical containers. The first bin supplies routine demand. When it is empty, the second bin is opened and a replenishment signal is sent. The second bin is intended to cover demand during replenishment lead time. When the delivery arrives, the bins are restored.

The two-bin method is a visual kanban-like control. It is effective for small, repetitive, low-value items with stable usage and standardized locations. It is less suitable for expensive, highly variable, perishable or serialized materials. A bin quantity should cover expected consumption during the replenishment period plus an appropriate buffer:

The approximation assumes fixed demand rate and lead time. In practice, bin size should be rounded to packaging, handling and supplier constraints. A major governance risk is that empty-bin signals may be ignored or that unauthorized stock is stored elsewhere, making the visual count unreliable.

8. Comparing Replenishment Systems

Selection should consider item criticality, demand variability, lead-time variability, value, shelf life, supplier minimums, transaction technology, shortage consequences and the cost of managerial attention. ABC classification alone is insufficient: a low-value item can be operationally critical, while a high-value item may have predictable demand and low shortage impact.

9. Cost, Service and Risk Design

The central trade-off is not simply “minimum inventory.” It is the economically and operationally appropriate balance between carrying cost and shortage risk. Relevant costs may include:

• ordering, setup and receiving cost;
• cycle-stock and safety-stock holding cost;
• purchase and transportation cost;
• shortage, expediting and lost-sales cost;
• production interruption and customer penalty cost;
• obsolescence, deterioration and disposal cost;
• information-system and control cost.

A service constraint can be expressed as a requirement that the selected policy achieve at least a target service level. Alternatively, a cost model may assign a shortage cost per unit. The latter is difficult because shortage cost can include intangible effects such as reputational damage and loss of future business.

Risk analysis should identify demand risk, lead-time risk, supply disruption, quality rejection, forecast bias, master-data error, geopolitical exposure and concentration of supply. Safety stock is not a substitute for supplier development, dual sourcing, quality assurance or realistic lead-time agreements.

A useful policy-selection matrix scores candidate policies against cost, service, risk, complexity and scalability. Weights should be agreed by operations, procurement, finance, production and customer-facing functions rather than selected solely by inventory control.

10. Parameter Governance and Implementation

A replenishment policy fails as often from bad data as from bad mathematics. Governance should include:

1. Clear ownership: a named owner for demand, lead time, parameters and exceptions.
2. Parameter segmentation: different rules for stable, intermittent, seasonal, critical and obsolete items.
3. Version control: documented effective dates and reasons for changes.
4. Exception management: escalation rules for stockout risk, excess stock and abnormal demand.
5. Supplier confirmation: actual lead-time distributions, minimum order quantities and order calendars.
6. Performance feedback: measurement of fill rate, CSL, inventory turns, aged stock, emergency orders and forecast error.
7. Periodic recalibration: parameters reviewed after structural changes, not blindly changed after every random fluctuation.

Enterprise resource planning, warehouse-management systems, barcode scanning, RFID, electronic data interchange and supplier portals can improve transaction visibility. Artificial intelligence can support demand sensing, lead-time prediction and anomaly detection, but it should not automatically change critical parameters without controls. Models require traceable data, monitoring for bias, human approval thresholds and an override process. A mathematically precise system using inaccurate receipts or unrecorded withdrawals will produce systematically wrong replenishment decisions.

11. Illustrative Case: MedEquip Components

Background

MedEquip, a fictional medical-device assembler, uses 1,800 sterile valve units per month. The supplier reviews orders weekly and has a nominal lead time of two weeks. Demand is moderately variable, supplier lead time occasionally extends to three weeks, and a shortage can stop a regulated assembly line. The existing policy is an informal monthly order based on visual inspection.

Problem

The plant experiences both excess inventory and emergency purchases. Records show that on-hand balances are not always updated when material is issued to production. Management wants better availability without doubling working capital.

Analysis

The first issue is not the choice between min–max and base-stock; it is data integrity. The team establishes barcode issue confirmation, separates quarantined stock from usable stock, and records open purchase orders in inventory position. Demand is approximately 90 units per working day, with standard deviation 25 units. The organization operates 20 working days per month. A weekly review is proposed, and the planning lead time is conservatively set at 15 working days to cover the review cycle plus supplier exposure.

Using the normal approximation, expected demand during protection is μ = 90 × 15 = 1,350 units. Standard deviation is σ = 25√15 ≈ 96.82 units. At a provisional 98% CSL, z ≈ 2.054, giving:

SS ≈ 2.054 × 96.82 ≈ 199 units.

The proposed periodic order-up-to level is therefore approximately S = 1,350 + 199 = 1,549 units, rounded to the supplier pack size. At each weekly review, the planner orders the positive difference between 1,549 and inventory position. Because lead time is itself variable, the team does not treat 1,549 as a permanent answer; it monitors actual service and lead-time distributions and tests the policy through scenario analysis.

Outcome and controls

After implementation, the plant reduces emergency orders and improves traceability. A monthly review examines CSL, fill rate, stock value, expiry exposure and supplier lead-time performance. If demand becomes seasonal or the supplier changes its schedule, the protection-period calculation is revised. The high service target is justified by line-stoppage consequences, but it is reviewed against expiry and working-capital cost.

Learning points

The case illustrates four principles: inventory position must be accurate; protection period must include the review interval; service measures must be defined; and safety stock cannot compensate indefinitely for unreliable supply or poor transaction discipline.

12. Challenges and Limitations

The normal approximation can misrepresent intermittent demand, demand with many zeros, promotional spikes and very short data histories. For intermittent items, Croston-type forecasting, compound distributions, empirical quantiles or simulation may be more suitable, although each has limitations. For highly seasonal items, parameters should be time-phased rather than based on a single annual average.

EOQ ignores many practical constraints. It may recommend an order quantity that violates supplier pack sizes, truck capacity, shelf life, budget limits or minimum order values. Likewise, a high calculated safety stock may be unacceptable for hazardous, controlled or perishable materials. In these circumstances, policy design is a constrained optimization or scenario-analysis problem rather than a single formula.

A further limitation is endogeneity: replenishment decisions can affect demand through availability, production scheduling and customer substitution. Historical demand observed during stockouts is censored and may underestimate true demand. Analysts should distinguish sales from requested demand whenever possible.

Best-Practice Design Sequence

A robust implementation sequence is:

1. Define the service objective and shortage consequence.
2. Clean item, supplier, lead-time and transaction data.
3. Segment items by demand pattern, criticality and supply risk.
4. Choose continuous or periodic review based on economics and capability.
5. Select the policy family: (s,Q), min–max, base-stock or two-bin.
6. Estimate demand and lead-time distributions using appropriate methods.
7. Set Q, ROP, S, B or bin quantities with documented assumptions.
8. Test scenarios for demand surges, delays, disruptions and parameter error.
9. Pilot the policy and measure service, cost and operational workload.
10. Govern exceptions and recalibrate after genuine structural change.

Chapter Summary

Replenishment systems convert inventory information into ordering action. Continuous-review policies respond quickly and commonly use an (s,Q) rule; periodic-review policies trade responsiveness for order consolidation and simpler review routines. Min–max policies are practical threshold systems, base-stock policies restore inventory position toward a target, and two-bin systems provide a visual control for routine items.

Order quantity, reorder point, review interval and order-up-to level address different decisions and should not be confused. EOQ provides a cost benchmark, while safety stock and reorder points protect against uncertainty. Correct design requires explicit service definitions, valid demand and lead-time assumptions, unit consistency, data governance and scenario testing. Cost, service, risk, criticality and operational capability must be evaluated together. Technology and AI can improve visibility and prediction, but governance, human accountability and reliable transaction data remain essential.

Key Terms

Inventory position: On-hand stock plus stock on order minus backorders.

Continuous review: Monitoring inventory position continuously or after each relevant transaction.

Periodic review: Assessing inventory at fixed time intervals.

Lead time: Time from order placement to usable receipt.

Protection period: Time demand must be covered before replenishment becomes available.

Cycle stock: Inventory held for expected demand between replenishments.

Safety stock: Buffer against demand, lead-time or supply uncertainty.

Reorder point: Inventory-position level that triggers an order.

Order-up-to level: Target inventory position after replenishment action.

EOQ: Economic order quantity balancing ordering and cycle-stock holding cost under stated assumptions.

(s,Q) policy: Policy ordering fixed quantity Q when position reaches threshold s.

Min–max policy: Policy ordering toward a maximum when inventory reaches a minimum.

Base-stock policy: Policy restoring inventory position toward a specified target.

Two-bin system: Visual replenishment system using a second container as the replenishment signal.

Cycle service level: Probability of no stockout during a replenishment cycle.

Fill rate: Proportion of demand supplied immediately from available stock.

Parameter governance: Controlled ownership, review and change of replenishment settings.

Intermittent demand: Demand occurring irregularly with periods of zero usage.

Review Questions

Short Answer Questions

1. What is the difference between on-hand inventory and inventory position?
2. Why does a periodic-review system require protection over review interval plus lead time?
3. Distinguish cycle service level from fill rate.
4. State two assumptions of the basic EOQ model.
5. Identify one appropriate and one inappropriate application of a two-bin system.

Long Answer Questions

1. Explain the design and operational differences between continuous-review and periodic-review systems.
2. Compare min–max, base-stock and (s,Q) policies, including their strengths and limitations.
3. Discuss how demand and lead-time uncertainty should influence safety-stock design.
4. Critically evaluate the usefulness and limitations of EOQ in a modern supply network.
5. Explain the data, governance and organizational conditions required for successful replenishment-policy implementation.

Analytical / Application Questions

1. Annual demand is 18,000 units, ordering cost is £60 per order and annual holding cost is £4 per unit. Calculate EOQ and explain two practical constraints that might require changing it.
2. Daily demand has mean 80 units and standard deviation 15 units. Lead time is fixed at 6 days. For a 95% CSL, calculate mean lead-time demand, safety stock and reorder point using z = 1.645. State the assumptions.
3. A supplier is reviewed every 7 days, lead time is 10 days, daily demand mean is 50 units and standard deviation is 8 units. Calculate a 98% order-up-to level using z = 2.054, and explain how variable lead time would affect the design.

Multiple Choice Questions with Answers

Question 1

Which quantity best represents inventory position?

A. On-hand inventory only

B. On-hand plus on-order inventory minus backorders

C. Annual demand minus annual supply

D. Safety stock plus cycle stock only

Correct Answer: B. On-hand plus on-order inventory minus backorders

Question 2

In a periodic-review system, the protection period normally includes:

A. Lead time only

B. Review interval only

C. Review interval plus lead time

D. Order quantity plus lead time

Correct Answer: C. Review interval plus lead time

Question 3

Under the basic EOQ assumptions, which two costs are balanced at the optimum?

A. Purchase and shortage costs

B. Ordering and cycle-stock holding costs

C. Transport and quality costs

D. Safety-stock and purchase costs

Correct Answer: B. Ordering and cycle-stock holding costs

Question 4

Annual demand is 12,000 units, ordering cost is £45 and annual holding cost is £3. What is the basic EOQ?

A. 200 units

B. 400 units

C. 600 units

D. 1,800 units

Correct Answer: C. 600 units

Question 5

Daily demand is 40 units, lead time is 5 days, and lead-time safety stock is 45 units. What is the reorder point?

A. 45 units

B. 200 units

C. 245 units

D. 400 units

Correct Answer: C. 245 units

Question 6

Which statement correctly describes a base-stock policy?

A. It orders only once per year

B. It restores inventory position toward a target level

C. It always orders a fixed quantity

D. It uses no inventory information

Correct Answer: B. It restores inventory position toward a target level

Question 7

A two-bin system is generally most suitable for:

A. Highly perishable, high-value items

B. Routine, low-value items with stable usage

C. Unique project equipment

D. Items with unknown demand and no replenishment lead time

Correct Answer: B. Routine, low-value items with stable usage

Question 8

If demand standard deviation per period is σ and fixed lead time is L periods, the standard deviation of lead-time demand under independent demand is:

A. σ/L

B. Lσ

C. √Lσ

D. L²σ

Correct Answer: C. √Lσ

Question 9

A 95% cycle service level is commonly associated with a standard-normal z value of approximately:

A. 0.645

B. 1.000

C. 1.645

D. 2.575

Correct Answer: C. 1.645

Question 10

Which is a major governance risk in replenishment systems?

A. Using consistent units

B. Recording receipts accurately

C. Allowing unrecorded stock movements to distort inventory position

D. Reviewing supplier lead time

Correct Answer: C. Allowing unrecorded stock movements to distort inventory position