AReCoS

Buyer's guide

Everyone saves 30%. How to tell which claims are real.

Heating, cooling and refrigeration vendors, AReCoS among them, all promise lower energy bills, and more of them every year say AI. This guide shows in pictures how these machines use energy, where each kind of product can change it, and the questions that separate a measured saving from a sales line. At the end we answer the same questions about ourselves.

1 · The physics in one picture

Cooling does not make cold. It moves heat.

An air conditioner, a heat pump, a chiller and a supermarket freezer are the same machine. A refrigerant boils in a cold coil and soaks up heat. A compressor squeezes the vapor until it is hotter than outdoors, and a hot coil gives that heat away. An expansion valve drops the pressure again, and the circuit repeats.

The refrigerant circuit, in cooling Warm room air flows into the cold indoor coil, which takes its heat and condenses water out of it. The refrigerant carries the heat to the compressor and on to the hot outdoor coil, which gives it to the outdoor air, then returns through the expansion valve. Compressor the energy you pay for Warm room air in Heat out Indoor coil cold: takes heat and water from the room air Outdoor coil hot: gives the heat to the outdoor air Water drains: dehumidifying Expansion valve
In cooling the indoor coil is cold and the outdoor coil hot. In winter a heat pump reverses the flow: the outdoor coil goes cold and pulls heat from the outdoor air, and that coil is where frost forms.

The compressor takes most of the electricity. Its work depends on the lift: how far it raises the temperature between the cold coil and the hot one. Every degree of lift saved is energy saved, so the fans, valves and pumps, which set the coil temperatures, matter as much as the compressor itself.

Dehumidifying is part of the job, and in humid places a big part

When room air meets a coil colder than its dew point, water condenses and drains away. Removing that water is the latent load; cooling the air itself is the sensible load. In humid climates the latent part can reach about 40% of the cooling load.

Cooling load in a humid climate: about 60% sensible, cooling the air, and up to about 40% latent, removing water. Sensible: cooling the air Latent: water about 60% up to about 40%
Typical of a humid climate; dry climates carry far less latent load. A saving claimed on the sensible part alone can be paid for in the latent part.

A control that saves energy by running the coil warmer removes less water, and can leave a building cool and damp. So ask whether humidity held, not only temperature.

Where we stand: dedicated dehumidification is not the AReCoS focus. Our controller is designed to run the whole unit, its coil included, inside the building's comfort limits. Specialist companies, such as Transaera, work on removing water more efficiently, and in a humid building they are worth a look alongside any heat pump or controller, ours included.

Heating is the same machine, reversed

A heat pump moves heat from outdoor air into a building, so it can deliver several units of heat for each unit of electricity. Below freezing, and above all in damp weather near freezing, its outdoor coil frosts. Section 6 explains why that matters to the bill.

2 · Where each product sits

Five layers, and only the bottom two move the machine.

Most energy-saving products are not in the machine at all. They sit above it and send it targets. That can be worth doing, but it limits what they can change. Ask any vendor which layer their product works in.

  1. Dashboards and analyticsShow data and flag faults. They save only when someone acts on them.
    Reads
  2. Building automationSchedules and setpoints: when to run, and what temperature to aim for.
    Sets targets
  3. Supervisory optimizers, often sold as AIMove setpoints through the day, from outside the unit, often from the cloud.
    Sets targets
  4. Unit controlsDrive the compressor, valves, fans and pumps. In most units, each on a loop of its own.
    Moves the machine
  5. HardwareCoils, compressor and heat exchanger: the best the machine can ever do.
    Sets the limit

AReCoS works in the bottom two: the controller in the unit, and our air–water heat exchanger.

A setpoint is a target, such as 55 °F supply air. The unit's own controls still decide how the compressor, valves and fans reach it.

The layers add up only in part. A better schedule and a better compressor setting save some of the same energy, so two products that each claim 20% do not make 40% together.

3 · Saves 30% of what?

The same saving can be 30%, 15% or 6%.

One saving, 6 units, three honest percentages.

Illustrative numbers, to show the arithmetic. Your building's split will differ.

Before comparing two claims, ask four things. A percentage of which energy: the compressor, the HVAC or the whole bill? Compared with what: the old unit, a new standard unit, or the same unit on its old settings? Over which weather: a mild month flatters almost anything. And measured or modeled?

How a real saving is measured

Measure a baseline on the same equipment before anything changes, then compare the same kind of hours afterwards, corrected for weather and load. The industry's methods for this are the International Performance Measurement and Verification Protocol (IPMVP) and ASHRAE Guideline 14. A vendor who uses one will tell you which option, and show you the baseline.

4 · Real, or just AI?

Nine questions to ask any vendor, including us.

  • What does it command?

    Name the actuators it moves. It optimizes with no answer usually means it sends setpoints, or only reports.

  • What was the baseline?

    The same equipment, measured before, corrected for weather. A comparison with an older unit is not a control saving.

  • Measured, or modeled?

    Both are useful; only one is a result. Ask for the site, the dates and the meter.

  • Did comfort and humidity hold?

    Running warmer, or longer between cycles, saves energy and can cost comfort, humidity or food quality.

  • What happens offline?

    If optimizing stops when the internet drops, so do the savings, and the control path runs through someone else's servers.

  • What about frost?

    A heat pump's costliest hours are cold and damp. Ask what the claim does in them, not in a mild month.

  • What does the AI predict, and change?

    A model is worth the decision it makes. Ask what it predicts, what it then changes, and what keeps it inside the machine's safeties.

  • Can you see the data?

    Interval data, before and after, from your own building or one like it, not only a chart in a slide.

  • Who checked it?

    Your own meter, a utility program or a third party, not the vendor's dashboard alone.

AI is not a bad sign in itself. Any controller that predicts is using a model, and ours is a physics model of the refrigerant cycle. The question is what the model is allowed to change.

5 · Why real time matters

The best setting changes all day, and every setting depends on the others.

The best compressor speed depends on the fan speed, which sets the coil temperatures, which depend on the valve opening, the weather, the load and the humidity, and all of those change through the day. In most machines each actuator has a control loop of its own: one holds a pressure, one a superheat, one a temperature. Each does its own job, none asks whether the whole machine is using the least power, and at times they work against each other.

One loop per actuator

Three separate loops: each sensor feeds its own small controller, which drives one actuator: compressor, valve or fans. loop loop loop Compressor Valve Fans Each fixes its own number

One decision for the unit

One controller: every sensor feeds a single controller running a model of the whole circuit, which sets the compressor, valve and fans together. One controller a model of the whole circuit, on the unit Compressor Valve Fans One decision, in real time
Left: how most units are controlled. Right: the AReCoS controller, designed to set the compressor, valves, fans and pumps together from one model of the whole refrigerant cycle, on the unit.

A setpoint sent from above, however clever, moves the target. It cannot change how the unit gets there. That is the layer we built for: one embedded controller, designed to estimate what its sensors cannot measure and to set every actuator together in real time, with no cloud in the control path. It is covered by U.S. Patent 12,768,557. Why setpoints alone fall short.

6 · Frost

In winter, the hours that cost the most are cold and damp.

When the outdoor coil of an air-source heat pump runs below freezing in humid air, frost grows on it. Airflow falls, heat output falls, and in the end the unit stops heating and reverses to melt the frost, with heat taken from the building. Backup heat, resistance strips or gas, usually covers that gap, and in many buildings that is where the winter bill goes.

A conventional defrost cycle Frost grows on the face of an outdoor coil, the airflow through it weakens and the heat delivered to the building falls; then the unit stops heating to defrost, and resumes once the frost has melted. A sketch, not data. Outdoor coil Heat to the building Frost grows, heat falls Defrost: heating stops Frost gone, heating again Backup heat often covers the gap A sketch of a typical cycle, not data
A conventional air-source unit: frost builds, output falls, and heating stops for each defrost.

So ask how often a unit defrosts, how long each defrost lasts, what heats the building meanwhile, and whether the savings claim was measured through those hours.

The AReCoS air–water heat exchanger takes a different approach. It is designed to control how much frost builds and to clear a thin layer while the unit keeps heating, with no reverse-cycle defrost in normal operation. If that system faults, the unit is designed to fall back to standard defrost and report it. Chamber tests are planned. How it clears frost.

7 · Our own answers

The same questions, answered about AReCoS.

We would rather be checked than believed. Here is where each answer stands today; the evidence page sorts every claim on this site the same way.

  • What does it command?

    Patented design
    The compressor, expansion valve, fans and pumps, designed to be set together in one decision on the unit. U.S. Patent 12,768,557.

  • What is the baseline?

    How an evaluation runs
    Your system as it runs today, measured by a method agreed up front, before the controller changes anything. The five steps.

  • Measured, or modeled?

    Modeled, and labeled so
    Our efficiency and savings figures come from models and design targets, and the site says so beside each one. The controller is under test in a research agreement (CRADA) with UT-Battelle, LLC, which manages Oak Ridge National Laboratory for the U.S. Department of Energy.

  • Do comfort and humidity hold?

    By design, within limits
    The controller is designed to optimize inside the building's comfort limits; a building automation system, where there is one, keeps those goals. Dedicated dehumidification is not our focus: in very humid buildings, look at the specialists too.

  • What happens offline?

    By design
    Designed to keep optimizing with the internet down. The cloud is optional, and only monitors.

  • What about frost?

    Patent-pending design
    Designed to keep heating without a reverse-cycle defrost stop in normal operation. Chamber tests are planned.

  • Who checks it?

    Your meter first
    Every evaluation is set against the baseline measured on your own system before anything goes further.

Ask us the hard questions.

Send us any question in this guide, about our controller, our heat pumps or a claim you have been offered, and we will answer it plainly.

Email info@arecos.net
625 Kenmoor Ave SE, Suite 350
Grand Rapids, Michigan 49546