> For the complete documentation index, see [llms.txt](https://docs.myrspoven.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.myrspoven.com/how-the-ai-saves-energy/indoor-climate-baseline.md).

# Indoor climate baseline

The indoor climate baseline is a set of comfort limits for indoor temperature that myCoreAI learns from the building's own history, from periods when Myrspoven was not steering. Where these limits are wider than the configured ones, myCoreAI uses them instead, which gives it more room to save energy. It is one of the mechanisms gated by AI Freedom.

## How the baseline is calculated

myCoreAI builds the baseline from **reference data**: indoor and outdoor temperature data from periods when Myrspoven was not steering the building. From that data it computes **reference limits**, a minimum and a maximum for each indoor temperature sensor that approximate the comfort bounds the building has run with.

* The minimum comes from winter data (October to March) and the maximum from summer data (May to September).
* Each season needs about six days of reference data. Without enough winter data, the configured minimum stays in place; without enough summer data, the configured maximum stays. Without either, myCoreAI uses the configured limits.
* Hours with unusually extreme outdoor temperatures and sensors stuck at one value are left out, and the limits come from the spread over many hours, so a single cold or warm hour does not skew them.

## What the baseline steers

The baseline only changes the limits of **indoor temperature sensors**. CO₂, supply temperatures and energy meters keep their configured limits.

myCoreAI keeps the average temperature of each sensor group, typically the rooms served by one heating circuit or air handling unit, between the average of its sensors' limits. A single room can sit outside its own limit while its group stays inside. See [Groupings](https://docs.myrspoven.com/how-the-ai-saves-energy/groupings.md).

## How the setting behaves

The baseline is controlled together with AI Freedom and is effectively binary:

| AI Freedom | Baseline                                               |
| ---------- | ------------------------------------------------------ |
| 0          | Off, unless the Indoor Climate Baseline setting is on. |
| > 0        | On. Any value above zero applies the full baseline.    |

## Which limit applies

When the baseline is active, myCoreAI compares the **configured limits** set in BackOffice with the **reference limits**, and the wider of the two applies. It judges this on the average across the building's sensors and the whole week, with office hours and other hours weighted by how many hours each covers.

|             | Reference limits wider on average                                                               | Configured limits wider on average                                                                                                         |
| ----------- | ----------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ |
| Lower bound | The reference minimum is used as it is, so the lower bound can be below the configured minimum. | The configured level applies: the reference minimum is shifted so its average matches the configured one, keeping its day and night shape. |
| Upper bound | The reference maximum is used as it is, so the upper bound can be above the configured maximum. | The configured level applies, with the reference maximum shifted the same way.                                                             |

On average, the baseline widens the comfort bounds or leaves them at the configured level. For individual sensors and hours, a limit can sit on either side of the configured value. Where a limit must hold, set AI Freedom to 0 and keep the Indoor Climate Baseline setting off.

## Office hours

myCoreAI separates office hours from the rest of the week, because tolerated comfort bounds often differ between workday and nights/weekends. Office hours are fixed for all buildings at **07:00–18:00, Monday to Friday**. Each reference limit therefore has one value for office hours and another for non-office hours. In a building with a consistent climate, the two values are nearly identical.

## Worked examples

**A building that ran warm.** An office building is configured with a lower limit of 20.5 °C. Its reference minimum is 21.5 °C during office hours and 20.5 °C at other times. The reference average is higher than the configured one, so myCoreAI keeps the reference shape and shifts it down until the average is 20.5 °C. The lower bound becomes 21.2 °C during office hours and 20.2 °C at other times: above the configured value in the daytime, below it at night.

**A building that ran cool.** Another building is configured with a lower limit of 20.0 °C, but its reference minimum averages 19.5 °C. The reference limits have the lower average, so myCoreAI uses them as they are and the lower bound averages 19.5 °C.

## When reference data is valid

Reference data only counts while it describes how the building runs today.

* **Vacancies:** a vacancy, or any other period that does not reflect normal use, is not valid reference data. Such periods can be marked so that they are left out.
* **Changes to the building:** after rebalancing, reprogramming of the building management system or a long shutdown, the earlier reference data no longer describes the building and needs to be renewed.

myCoreAI does not detect these situations by itself. Tell your Myrspoven contact when they happen.

## How reference data is collected

Reference data is also what savings are measured against (see [Climate compensation](https://docs.myrspoven.com/measure-savings/climate-compensated.md)). It comes from two kinds of periods:

* **Before the start:** until myCoreAI starts steering, the building runs on its own control, so data collected from the moment the building is connected counts.
* **Measurement weeks:** during the contract, the only way to collect reference data is to switch optimization off. When a building's reference data is missing a range of outdoor temperatures, Myrspoven switches optimization off for about a week while the weather covers that range. The week is chosen from the missing range and the forecast, not from the calendar. Depending on your agreement, you approve the week beforehand or are told before it starts.

A shorter window does not work, because the building needs one to two days to settle at its own temperature after optimization is switched off. New reference data adds to what the building already has, so a building whose history already covers its outdoor temperatures needs no measurement week.

## The Climate Baseline chart line is not a limit

Group Control in myPortal 2.0 has a chart line called **Climate Baseline**. It shows the indoor temperature the group would be expected to have without Myrspoven: a curve fitted to the group's average indoor temperature against outdoor temperature on days when Myrspoven was not steering. It is there for comparison. myCoreAI does not steer towards it, and it is not one of the limits on this page.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.myrspoven.com/how-the-ai-saves-energy/indoor-climate-baseline.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
