Analysis of Continuous-Review Inventory Policy Service Level in Intermittent Demand Environments

Abstract

This paper investigates the reliability of the continuous-review (R, Q) inventory policy in environments characterized by extreme demand intermittency. Using a 23-period demand trace from a fishery diagnostic laboratory with a Coefficient of Variation (CV) of 2.08 and a 61% zero-demand rate, we stress-test the classical "Normality Assumption." While reorder points (R) are traditionally calculated using Gaussian safety stock formulas to meet target service levels, we hypothesize that this approach suffers from Mathematical Decay when subjected to skewed, lumpy demand.Using a dual-track methodology, we first perform a Static Stress Test (non-parametric bootstrap) to isolate the reorder point. Results show a significant service-level deficit, where the "Normal" R fails to cover empirical demand spikes, falling nearly 11% below the 98% target. We then conduct a Dynamic System Simulation to observe the interaction between $R$ and the order quantity (Q). This reveal a phenomenon we define as the "Masking Effect": as Q increases, the system’s service level recovers to near-target levels despite using the same faulty reorder trigger.The study concludes that in intermittent environments, the reorder point is a decoupled and unreliable trigger; system survival depends almost entirely on the "brute force" of the replenishment volume. These findings suggest that practitioners in specialized sectors should move away from parametric safety stock math in favor of percentile-based empirical triggers to avoid the hidden operational risks created by the masking effect.