Advanced process control took four decades to reach the cold chain because the arithmetic did not work until recently. The technique came out of oil refining in the 1970s and 1980s, where a fraction of a percent of yield improvement paid for the engineering many times over. A refrigerated warehouse had no yield to optimize, and its energy bill never justified a bespoke control project per site.
Where the technique came from
Basic process control holds a value at a setpoint. Advanced process control sits above it, deciding what the setpoints should be by running a model of how the process responds and choosing adjustments that best satisfy an objective across interacting variables at once.
Refineries had ideal conditions: continuous processes, expensive feedstock, tightly coupled units, and an objective that translated into dollars per barrel. Cold storage had none of that. The process is simpler, the objective is to hold a temperature, and the facilities are numerous and modest.
What changed the math
Three shifts moved the calculation:
- Energy became a more volatile and more visible cost.
- Computing became cheap enough that the control layer no longer required a dedicated engineer per installation.
- Software could be written once and deployed across many sites instead of rebuilt at each.
That third shift carries the argument. Enterprise control platforms built for refrigerated portfolios, among them CrossnoKaye, install software above the existing equipment controls at each site, so a facility keeps its compressors, panels, and vendor systems while gaining a common way to operate them.
Two refrigerated warehouses are never identical, but they share structure: the same equipment categories, the same relationships, the same physics. A model that captures that structure once can be fitted to a site through configuration rather than reengineering, so the marginal cost of reaching the fortieth facility approaches zero. A benefit too small to justify one project becomes substantial across a portfolio.
Public funding sharpened the case
California's Food Production Investment Program has awarded $137 million to 63 projects, with projected annual savings of 16 million kilowatt-hours and 25 million therms. Grants like that also set measurement expectations, which surface what control improvements are worth.
Rate design points the same way. A December 2025 ACEEE analysis identifies rate improvements that would push power use of new electric equipment toward nonpeak hours. For a refrigerated facility, shifting load off peak is exactly the multi-variable timing decision advanced control is built to make.
The adoption gap is still real
A 2025 review of automated building climate control shows how far field evidence sits from enthusiasm.
| Finding | Detail |
|---|---|
| Scope | 104 papers on model predictive control and reinforcement learning |
| Industry adoption | Neither approach widely adopted |
| Field demonstrations | 71 percent used protocols that may produce unreliable estimates |
| Reliable demonstrations | 16 percent average savings residential, 13 percent commercial |
A savings figure means little without a stated baseline, a measurement period, and an account of who verified it.
California as the forcing case
The state concentrates every variable that makes the case:
- USDA's biennial survey, published February 2026, put national gross refrigerated capacity at 3.99 billion cubic feet across 931 warehouses, California holding the largest share at 400 million.
- An industrial base working through electrification, and rate structures that reward operators able to move load.
- A food processing sector with public funding tied to measured outcomes.
Four decades is a long delay for a technique never in doubt technically. It waited for the cost of applying it to fall below the value, across enough sites to notice. Both conditions now hold.
Copyright © 2026 California Business Journal. All Rights Reserved.
For California Business Journal Disclaimers, go to /terms-conditions/.
