The Price of Simplicity: Analyzing Decoupled Policies for Multi-Location Inventory Control
Yohan John*, Vade Shah*, James A. Preiss, Mahnoosh Alizadeh, Jason R. Marden · 2025
Read The Price of Simplicity: Analyzing Decoupled Policies for Multi-Location Inventory Control by Yohan John*, Vade Shah*, James A. Preiss, Mahnoosh Alizadeh, Jason R. Marden (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-22639.txt.
Introduction
Yohan John, Vade Shah, James Preiss, Mahnoosh Alizadeh, and Jason Marden study simple control rules for stock kept at many sites, in work from 2025. A planner must meet random demand at each site. Costs are linked because buying in bulk can save money across sites.
The team asks what is lost when each site is run on its own, in a simple way. They show a surprise result about the best stock levels for single sites. They also give tight bounds on how much worse simple rules can be. The gap depends mainly on how strong the bulk savings are.
Catalog record
| CREATOR | Yohan John*, Vade Shah*, James A. Preiss, Mahnoosh Alizadeh, Jason R. Marden |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-22639
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Price of Simplicity: Analyzing Decoupled Policies for Multi-Location Inventory Control Yohan John*, Vade Shah*, James A. Preiss, Mahnoosh Alizadeh, Jason R. Marden Abstract What is the performance cost of using simple, decoupled control policies in inherently coupled systems? Motivated by industrial refrigeration systems, where centralized compressors exhibit economies of scale yet traditional control employs decoupled room-by-room temperature regulation, we address this question through the lens of multi-location inventory control. Here, a planner manages multiple inventories to meet stochastic demand while minimizing costs that are coupled through nonlinear ordering functions reflecting economies of scale. Our main contributions are: (i) a surprising equivalence result showing that optimal stationary base-stock levels for individual locations remain unchanged despite the coupling when restricting attention to decoupled strategies; (ii) tight performance bounds for simple decoupled policies relative to optimal coupled policies, revealing that the worst-case ratio depends primarily on the degree of nonlinearity in the cost function and scales with the number of locations for systems with fixed costs; and (iii) analysis of practical online algorithms that achieve competitive performance without solving complex dynamic programs. Numerical simulations demonstrate that while decoupled policies significantly outperform their worst-case guarantees in typical scenarios, they still exhibit meaningful suboptimality compared to fully coordinated strategies. These results provide actionable guidance for system operators navigating the trade-off between control complexity and operational efficiency in coupled systems. I Introduction Industrial refrigeration poses significant control challenges, particularly in cold storage facilities where multiple rooms must be maintained below critical temperature thresholds to preserve perishable goods. These systems use evaporators in each room that absorb heat from storage spaces and a centralized set of compressors—the dominant energy consumers—to meet the aggregate cooling load. Conventional control approaches rely on simple temperature setpoints that determine whether individual evaporators are on or off, creating a decoupled system where evaporator operation depends solely on conditions within each room [ 13 , 24 , 9 ] . However, cooling multiple rooms simultaneously could be more energy efficient because compressors operate most efficiently at maximum capacity [ 19 , 21 ] . Although technologies now exist to enable such coordinated control strategies [ 5 ] , it is crucial to determine whether the resulting efficiency gains would justify the implementation costs. Cold storage facilities can be thought of as multi-location inventory control systems [ 15 , 2 , 10 ] , a model from operations research where a planner manages multiple inventories of an item that is depleted by a random demand process over time; the core challenge is to balance the competing costs of holding excess inventory, incurring backlogs when demand exceeds the current inventory level, and placing orders for additional inventory. In this framework, cooling capacity corresponds to the ‘inventory level’ and heat loads represent ‘demands’ that deplete this inventory. The ‘thermal battery’ concept, i.e., intentionally overcooling products below required thresholds, parallels holding excess stock in traditional inventory systems, insufficient cooling is akin to backlogs, and the energy consumed by cooling operations represents ordering. When multi-location inventory systems are decoupled with independent demands and costs, they reduce to separate single-location systems, and decoupled control policies (like the setpoint control approaches used in refrigeration systems) are often effective. However, when the systems are coupled through their costs, as in refrigeration systems where compressor efficiency creates interdependencies, the problem and its optimal solutions may become more complex. Inventory control provides a well-established framework for studying coupling in these kinds of systems. Decades of research on the problem have focused on characterizing optimal policies under various assumptions on the demand process and cost functions for both single-location and multi-location systems. The key insight from this literature is that for typical single-location problems, simple control policies such as base-stock policies (order up to a specific threshold S if inventory falls below S ) or (s,S) policies (order up to S if inventory falls below s ) are provably optimal [ 1 , 22 ] . The multi-location inventory problem generalizes the standard problem to multiple locations. When the systems are decoupled, one can achieve optimal or near-optimal performance with decoupled policies [ 1 , 15 ] . However, in coupled systems where the cost in one location depends on another, such as when per-unit costs decrease with larger order quantities [ 20 , 3 ] , the problem is no longer decomposable, and simple decoupled policies may be significantly suboptimal. Several works have characterized optimal [ 2 , 15 , 7 , 8 , 18 ] or near-optimal [ 12 , 10 , 11 ] policies for various kinds of coupled multi-location systems. Importantly, these policies can be complex and difficult to implement in practice. The focus of this work is to study how well simple, decoupled policies perform in coupled multi-location inventory systems like the refrigeration example above. In particular, we focus on systems where demand and holding/backlog costs are independent across locations, but ordering costs are a nonlinear function of the cumulative order quantity across all locations in each period. Our results are summarized as follows: (i) Our first contribution given in Theorem 1 demonstrates that the optimal stationary base-stock policy for a particular location in a multi-location inventory control problem is equivalent to the optimal stationary base-stock policy for that location in isolation. In other words, despite the coupling through nonlinear ordering costs, the existence of other locations does not influence the optimal base-stock level for a given location. This result significantly simplifies the process of finding optimal stationary base-stock policies, as it allows us to solve separate single-location problems rather than a joint multi-location problem. (ii) Our second contribution given in Theorem 2 provides tight bounds on the performance gap between the optimal policy and the optimal decoupled base-stock and (s,S) policies in multi-location inventory control problems. Specifically, for sector-bounded ordering costs, we show that this ratio depends only on the degree of nonlinearity of the ordering cost function, while for affine-bounded ordering costs, it also scales with the number of locations. These bounds precisely quantify the trade-off between policy simplicity and performance across different system structures, providing guidance on when decoupled base-stock policies are nearly optimal and when more complex coupled policies may be necessary. (iii) Our third contribution extends our analysis to a simple online policy as an alternative to optimal decoupled base-stock or (s,S) policies, which are structurally simple but can be computationally difficult to identify, especially in the case where demands are correlated over time [ 6 ] . In Proposition 3 , we analyze the randomized cost-balancing policy [ 17 ] , an efficiently computable heuristic that balances ordering, holding, and backlog costs. We prove that this algorithm achieves competitive ratios that scale predictably with our bounds for optimal decoupled policies, thereby demonstrating that effective inventory control is possible even with online decision rules. These results are particularly valuable for practical applications where computational resources are limited or real-time decisions are required. We validate these theoretical results through numerical simulations which reveal that while simple decoupled policies and online algorithms outperform their worst-case bounds in practice, they still exhibit significant suboptimality compared to optimal coupled policies. Although our motivation stems from industrial refrigeration systems, we recognize that real refrigeration dynamics involve additional complexities