How a Hidden Default Curve Turned Into +7% More Savings (Chiller Curves + Sequencing)

This week I reviewed an energy model that looked fine at first glance.

Reports were clean. 23% Savings looked decent. No obvious red flags.

But once I opened the plant, I found something that shows up more often than most teams realize:

The model was running on a default chiller curve.

And that one “quiet” default turned into a meaningful result once we corrected it — and then took the review one layer deeper.

The hidden issue: default chiller curves

Most modeling software will let you run a chiller plant using generic/default performance curves.

The model doesn’t crash. The reports don’t complain. The project team moves forward.

The problem is that chillers rarely spend most of their operating hours at full load. A large portion of annual energy is accumulated at part-load, where the curve shape matters a lot.

So when the curve is generic, the energy is generic, and the savings can be distorted in either direction.

In this case, the default curve was undercounting the savings that the actual chiller could deliver.

Step 1 - Replace the curve: +4% improvement

First, we replaced the default curve with the official manufacturer performance data (normalized for modeling).

That alone improved the model results by about 4%.

This is the point where many teams stop and take the win.

• model runs ✅
• savings improved ✅
• reports still look normal ✅

But the truth is: correcting curves often reveals a second layer.

Because once the equipment performance is realistic… controls matter more.

Default vs Manufacturer Curve (Illustration)

(Illustration: default curve vs manufacturer curve — normalized. Not an exact submittal.)

Even in a simplified view, you can see the point:

Curve shape drives part-load energy. And part-load drives annual energy.

Step 2 - Sequencing and staging: the extra +3%

After the curve swap, we checked one more thing:

Sequencing / staging logic.

This is where a lot of “plant savings” fall apart, not because the equipment is wrong, but because the plant is effectively being simulated as if it’s operating in a way that would never happen in real life.

Once we updated sequencing to match realistic operation at part load, the results improved again.

Total outcome after both changes:

✅ 30% savings over the Baseline  (+~7% improvement)

That’s the two-step pattern I see constantly:

• Curves get you partway there
• Controls and sequencing unlock the rest

Why this matters (and why it affects review)

Plant modeling issues are dangerous because they often:

• don’t show up in the standard reports
• don’t cause simulation errors
• don’t look “wrong” unless you open the guts of the file

But reviewers (and experienced QA reviewers) know where these mistakes hide.

And even when the reviewer doesn’t call it out directly, fragile plant assumptions often collapse under scrutiny when:

• someone asks for part load justification
• results don’t pass basic sanity checks
• the part-load behavior doesn’t match the narrative

The goal isn’t to “optimize a model.”

The goal is to produce savings that are both:

1. realistic, and
2. defensible / reviewer-safe

The bigger takeaway

If you’re modeling chillers, these are some of the highest-value areas to QA:

• Are you using default curves anywhere in the plant?
• Do the curves reflect actual equipment performance?
• Does sequencing reflect realistic staging behavior?
• Do part-load results pass simple sanity checks?

The trap is that everything can look fine… right up until someone experienced opens the plant.

Want a second set of eyes before submission?

This exact issue is what our Second-Look Model Review is designed to catch:
defaults and plant assumptions that don’t show up in the reports but quietly drive the outcome.

If you’d like a targeted plant QA review (or help tightening your modeling assumptions to be reviewer-safe), reach out here:

Contact / Freelance Modeling