Energy planning

The hidden cost of a suboptimal energy design

The most expensive decisions in an energy project are made early, on little data, and then locked in for 20 to 30 years. When the design is wrong, nothing breaks. The plant runs, the bills get paid, and nobody ever sees the more efficient and more reliable system that was never built.

See what the audits found ↓

Why the cost stays hidden

A badly chosen system doesn't fail. It runs.

An oversized boiler still delivers heat. A heat pump sized on the wrong demand still keeps the building warm. The extra cost shows up as slightly higher bills and capital that earns less than it could, every year, and no invoice ever names it.

Most plants are never checked against what they were meant to achieve. When Bavaria's state audit office reviewed combined heat and power (CHP) units in state buildings in 2025, it found that in one case a unit planned for 5,000 full-load hours a year was running about 1,140. Eleven of the economic calculations behind these units had badly overestimated how much they would run.

And these are public buildings, where an audit office eventually looks. Most private and municipal sites never get that review.

Looking back

Where someone took the time to look

Most energy systems are never checked against the plan they were built on. Now and then an audit office goes back after the fact and compares the decision with what actually happened. Here are two of those reviews, one in Germany and one in the US.

Biberach: a heating plant twice the size it needed

€90,000 a year

that a conventional design would have saved, according to the state audit office. Around €1.8 million over a 20-year plant life (our estimate).

A police college in Biberach, Germany, got an innovative energy centre in 2015 for €2.4 million: CHP, a heat pump, an electric heater, solar thermal and heat and cold storage, designed to sell flexibility to the power grid. The external consultant had advised against the concept as economically inadvisable.

The Baden-Württemberg audit office later found that heat use came in more than 500 MWh a year below plan. The CHP ran only 2,200 full-load hours, under 25% utilisation. The electric heater and solar thermal had no effect at all, and the heating centre was 100% oversized.

US Capitol Power Plant: $85 million on one price outlook

$85 million

total project cost committed on an analysis the agency's own Inspector General later said may have overstated the benefits.

In Washington, the Architect of the Capitol chose a 7.5 MW gas-turbine cogeneration plant: about $57 million in construction, $24 million in financing and $4 million in project management. The business case promised about $7.3 million in savings over 27 years. A plain gas boiler, at $9.3 million, would have saved about $2.7 million.

The US Government Accountability Office found that the agency "did not perform valid sensitivity or uncertainty analyses": gas prices were varied by a flat 25% with no rationale given. In 2019 the Inspector General concluded the analysis "may have overstated the financial benefits", partly because financing charges were understated and extra maintenance costs were not consistently included.

The common thread

Not a lack of expertise. A lack of options compared.

None of these projects lacked engineers or consultants. The audit reports don't tell us everything about how each study was done, but they point to familiar limits of conventional planning:

A limited set of variants. Every option worked through by hand takes time, so studies typically compare a handful of options chosen at the outset.
One main price outlook. At the Capitol, prices were varied by a flat 25% with no rationale. More often than not, options are ranked mainly on one expected price path.
Demand and sizing assumptions that go unchallenged. At Biberach, heat use came in more than 500 MWh a year below plan. In Bavaria, eleven calculations overestimated running hours. When inputs aren't tested against a range, oversizing goes unnoticed until the plant is running.

A broader analysis was possible in most of these cases. But when each extra scenario means another round of consultant work, it is easy to see why it often doesn't happen.

What an optimiser does differently

Not "how does this design perform?" but "which design performs best?"

A simulation tool tells you how a design you chose performs. An optimiser searches the combinations of technologies and capacities for you, instead of evaluating only the few you thought of.

Every hour of the year.

Demand, solar yield, prices and storage are modelled across all 8,760 hours, so sizing reflects how the system actually runs.

Many options at once.

Heat pumps, CHP, boilers, PV, batteries, thermal storage, networks and grid connections are weighed together, not one at a time.

Many futures.

Re-running the analysis under high and low energy prices, CO₂ prices or demand takes hours, not another study. You see which design stays good across the range, not just which one wins on the forecast.

Trade-offs made visible.

Cost against CO₂ against self-sufficiency, so the decision is an informed choice rather than a single number.

None of this replaces engineering judgement. It gives that judgement far more to work with, early, while changes are still cheap.

Not the best of the three options you had time to compare. The best of thousands.

That is what Sympheny does.

What an optimiser can't do

Good engineers with better tools

An optimiser is only as good as its inputs. At the Suurstoffi district in Rotkreuz, monitoring showed heating demand at twice the planning calculation, and by spring 2014 the borehole field was completely discharged. Optimising on the wrong demand would only have produced a confidently wrong answer. Spotting that a demand figure doesn't fit the building type is engineering judgement, not computation.

At Biberach, the consultant's warning was right, and the project went ahead anyway. A warning backed by a full comparison of variants and futures is much harder to overrule than a professional opinion.

That is the division of labour Sympheny is built for. The optimiser does the exhaustive part: thousands of combinations, every hour of the year, every scenario. The engineer does what no solver can: frame the question, judge the inputs, and stand behind the recommendation. Every result traces back to its inputs, so engineers can question it rather than take it on trust. And when a modelling question gets tricky, our team of PhD specialists in multi-energy optimisation, with more than ten years of research at Empa behind them, is there to back you up.

Building or reviewing a model yourself? Five checks before you trust an energy model →

What it looks like in practice

Port of Switzerland, Basel: planning against a range, not a point

At Switzerland's only commercial port, IWB and the port's partners wanted to know what to do with a large PV surplus: export it, share it, store it, or something else. The site has many buildings and operators, very different loads, and an energy market that shifts often.

IWB used Sympheny to compare four strategies across 16 scenario variants before committing to anything. All strategies other than the status quo cut costs by up to 20–25%. The analysis also showed that regulatory change, not market price volatility, is the dominant risk, and that interlinking the buildings is the strategy that protects the port against a negative tariff change.

Read the Port of Switzerland case study →

Sources

Planning a project?

Talk to us before the variants are fixed.

If you are about to commit to an energy concept for a district, campus, port or industrial site, we'll show you what an optimised analysis of your project would look like, and where the money is.

Book a demo Start a free trial