Conservant Systems · July 2026 · 3-minute read
Efficiency claims have a credibility problem, and everyone in the industry knows why: baselines drift, weather varies, occupancy changes, and a determined analyst can make almost any retrofit look good against the right reference year. The gold standard rebuttal is an interrupted time-series on the same building — turn the measure off, watch the meters, turn it back on.
That's exactly what happened at the Timken Museum of Art in San Diego, where NREL ran instrumented testing across three seasons (reports TP-5500-90261 and TP-5500-94847, alongside DOE-HIT, GSA Green Proving Ground and CalNEXT evaluations). For one week, the HEDS energy-recovery and optimization functions were disabled and the building ran on its conventional logic.
+87% electrical. +129% fossil fuel. Same building, same collection-grade humidity requirements, same week of San Diego weather — with only the technology switched off.
Why this test is hard to argue with
- The building is its own baseline. No reference-year modeling, no weather normalization debates — the counterfactual ran live, in the same envelope, serving the same loads.
- The duty is unforgiving. A fine-arts museum holds tight temperature and RH around the clock to protect irreplaceable work. Nobody relaxed the requirements to flatter the numbers; conditions stayed in spec throughout the test period under both modes.
- Third parties held the meters. NREL led the instrumentation; CalNEXT — the California IOUs' independent emerging-technology program — separately evaluated the site, found annual savings above 75%, and recommended the state's utilities build incentive programs for HEDS.
The numbers around the test
Across operating modes, NREL documented 39–72% electrical savings and 64–99% fossil-fuel reductions at the Timken; CalNEXT's full-year extrapolation landed at 79% electricity and 91% gas. The site was also converted from constant-air-volume reheat (as it had operated for ~60 years) to HEDS energy-recovery VAV — which is where the top of that savings band comes from. The dataset now serves as a primary calibration source for the government-backed EnergyPlus HEDS module NREL is building.
For M&V professionals, underwriters and skeptical engineers, the off-switch week is the shortcut: whatever you believe about models, the meters moved when the switch did.
Want the source documents? Every report referenced here is available — and the fastest way to check the present tense is to watch a live project during a call with the inventor.