A hardware replacement and a software optimization do not compete on equal terms. New equipment can improve efficiency, but optimizing the software layer controlling existing equipment can return value sooner, without construction. The hardware case remains valid. The missed calculation is failing to assess that operational improvement first.
Comparing Capital Efficiency
Replacing a chiller in a mid-size commercial building generally costs $150,000 to $400,000, based on capacity and complexity. Savings come from better efficiency: a magnetic bearing centrifugal chiller at 0.35 kW/ton versus an older centrifugal unit at 0.65-0.75 kW/ton delivers a meaningful full-load gain. That gain is real, and replacement makes sense when the equipment is nearing end of life.
The calculation is roughly $200,000 in capital cost against $18,000-25,000 in yearly energy savings, depending on hours at load and rate structure. Simple payback: 8-11 years. Aging-equipment maintenance costs can make that a sound investment with a sound return.
Now put that new chiller under control logic unchanged since commissioning. Sequencing starts the second chiller when return water passes 56F, without regard to load distribution or thermal conditions. Supply temperature stays fixed at 44F year-round, even when 48F or 50F would serve the building on mild days. At full load, the new chiller delivers 0.35 kW/ton. On mild days at part load with that fixed setpoint, performance falls well below it. The controls therefore leave some hardware efficiency gains unused.
The Software ROI Case
Software optimization has another cost profile: no equipment capital outlay, but subscription software plus an integration setup fee, typically $8,000-15,000 for initial BMS connection and commissioning. Annual operating cost is $15,000-40,000, based on building count and complexity tier.
In the same building, continuous setpoint optimization and occupancy scheduling typically cut HVAC energy cost by 10-18%. If electricity costs $180,000 per year and HVAC accounts for 50-60%, a 12% HVAC improvement saves $10,800-12,960 annually in direct energy costs. Peak shaving can also reduce demand charges, although the amount depends heavily on the utility rate structure.
Software payback is 14-22 months. That does not make hardware a poor investment. It clarifies order of operations. The software layer returns sooner and can sharpen the hardware decision by showing where efficiency gains actually originate before capital is committed.
The Calculus Directors Misread
The frequent mistake is testing hardware and software as separate choices instead of in sequence. A chiller under optimized controls outperforms the same machine using commissioning-day settings. A specification-sheet rating reflects laboratory conditions. In the field, results depend on how controls respond to the building's actual conditions.
For a chiller replacement, compare a new chiller with optimized controls against the existing chiller with optimized controls. Comparing both machines under current controls obscures the result. Better operation can sometimes defer replacement of sound existing equipment by three to five years while providing comparable performance at a fraction of the cost.
Another error is assigning to hardware savings that arrived alongside operational changes. A single project may replace chillers, rebalance the hydronic system, and add occupancy sensors, then credit every reduction to the equipment. Operational changes may have produced half or more of the total savings. That leaves later hardware-only decisions measured against a baseline that is too favorable.
Investment Sequence
The highest-return sequence begins with software. Connect BMS data to an optimization engine, operate it for 60-90 days, and measure operational savings alone. That record provides several useful answers.
It creates a measured baseline tied to actual occupancy and equipment condition, rather than relying only on an energy model from commissioning. It shows the largest remaining gaps. If 70% of avoidable waste comes from inefficient HVAC schedules, occupancy-based scheduling comes first. If 70% comes from part-load chiller plant inefficiency, sequencing optimization takes priority. It also separates equipment near the edge of its efficiency curve, where replacement may help, from equipment already performing well.
Once the software layer is operating, later hardware purchases can earn more. With supply air temperature reset, chilled water temperature reset, and continuous sequencing already active, a new chiller should remain near its rated efficiency for more operating hours than one installed under commissioning-day settings.
Life-Safety and Reliability Matter
This does not mean end-of-life equipment should run forever under optimized controls. Reliability or life-safety risk can require replacement regardless of software. A repair-heavy aging chiller, a structurally compromised cooling tower, or AHUs using motors beyond their service life calls for a continuity and safety decision, not an energy-only calculation.
The narrower point concerns mechanically sound, reliable equipment: begin with software optimization rather than replacement. Where replacement is warranted, running the software first gives the decision a measured baseline and a clearer view of the likely efficiency gains.