Monitoring-based commissioning: keeping savings from fading

Most retro-commissioning projects work. A team tunes the existing systems, the meter drops, and everyone signs off. Then, over the next three to five years, setpoints drift, overrides accumulate, a controller is replaced, and the savings quietly leak away. Monitoring-based commissioning (MBCx) exists to stop that leak by turning a one-time tune into a continuous process fed by meter data.

MBCx versus one-time retro-commissioning

Retro-commissioning, also called existing building commissioning, is a periodic investigation of how existing systems are actually running, followed by low and no-cost corrections. Natural Resources Canada reports that existing building commissioning typically delivers 5% to 20% energy savings with a payback of three years or less. The weakness is persistence. Without continuous measurement, no one knows when the savings start to erode.

MBCx adds permanent metering and analytics so faults are caught as they reappear, not at the next audit years later. It is worth separating from two adjacent practices. An ASHRAE-style energy audit identifies opportunities at a point in time; MBCx is an ongoing operational process. And measurement and verification under IPMVP quantifies savings against a baseline, while MBCx generates and sustains them. Good MBCx borrows M&V-grade baselines to prove that the savings are real.

What the numbers say

The largest dataset on this comes from an LBNL and BCxA study of about 1,500 North American buildings. It found median existing-building savings of 6.4%, with monitoring-based commissioning under utility programs reaching a median 9%, and projects run outside utility programs reaching 14%. Median existing-building commissioning cost was about $0.26 per square foot, and the median simple payback was 1.7 years, with a 25th to 75th percentile range of 0.8 to 3.5 years. Separately, the Smart Energy Analytics Campaign found buildings using energy information systems with fault detection reached a median 9% saving with a one to two year payback. The extra metering pays for itself and the savings last longer.

One finding worth stressing is that these savings tend to grow, not fade, when the monitoring is genuinely continuous. A one-time project captures a step change and then decays. A monitored building keeps catching the next fault, so the savings curve trends the right way over years rather than sliding back toward the baseline. That persistence, more than the size of the first-year saving, is the real argument for the extra metering.

The data pipeline MBCx needs

MBCx is only as strong as the data behind it. A workable pipeline has a few non-negotiable parts:

• Permanent interval metering at the whole-building level, plus submeters on major systems where they exist
• Automated, reliable data collection so gaps do not corrupt the trend
• Clean, standardized, unit-consistent series across every commodity and meter
• Weather data and a normalization method so savings are not confused with a mild winter
• A baseline model good enough to defend savings claims
• Fault rules and analytics that turn the data into named problems
• A workflow to assign, fix, and verify each issue

The step that quietly makes or breaks MBCx is the third one. If meters are relabelled after tenant changes, if bills span uneven periods, or if hourly and monthly data are mixed without reconciliation, the analytics produce noise. Standardizing bills and interval feeds into analysis-ready data is the foundation, not an afterthought.

The baseline deserves particular care, because every savings claim is measured against it. A baseline built on a single unusual year, or one that ignores a change in occupancy or operating hours, will over- or under-state savings for the life of the program. Sound practice is to build the baseline from a representative period, adjust it for weather and for known changes in how the building is used, and revisit it when the building itself changes materially. That is where MBCx and measurement and verification meet: the same normalized baseline that proves a saving is also the reference the analytics watch for drift.

What MBCx does and does not fix

MBCx is an operational discipline, not a capital program. It excels at catching control and scheduling faults, drift in setpoints, and equipment left running when it should not be. It will not replace an undersized chiller, reinsulate a leaky envelope, or fix a fundamentally oversized system. Those are retrofit questions. The value of running MBCx first is that it tells you which of your remaining problems are operational, and therefore nearly free to solve, before you spend capital. A building that has never been continuously monitored often finds that a meaningful share of its excess consumption disappears with no capital at all, which sharpens the business case for the retrofits that remain.

Who runs it, and how often they look

The technology is the easy part. MBCx works when a named person or team reviews the analytics on a regular cadence, weekly at first, then at whatever interval keeps the fault queue short, and has the authority to get issues fixed. Without that loop, an analytics platform becomes another dashboard nobody opens. The most effective programs treat the review as a standing operational meeting, tie each fault to an owner and a due date, and track the resulting savings against the baseline so the effort stays visible to the people who fund it.

Metering granularity is a design choice worth making deliberately. Whole-building interval data is enough to see that a building is drifting; submeters on major systems are what let you say which system. More meters mean more insight and more cost, so a common pattern is to start at the whole-building level, then add submetering on the largest and least understood loads once the analytics show where the uncertainty is.

Getting started in a Canadian building

In Canada, the natural on-ramp is the existing building commissioning framework, which NRCan describes as ongoing commissioning that uses technology and software to monitor and optimize operations. Start with a retro-commissioning pass to fix the obvious faults, install or expose the metering you need, then keep the analytics running so the next fault is caught in weeks rather than years.