Skip to content
Permalink
3b68abe080
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Go to file
 
 
Cannot retrieve contributors at this time

A/B Testing and optimisation for heat pump-based heating in buildings

Project summary

During rapid development of a heat pump-based heating system for a residential building, there is the need to compare configuration A with configuration B. A configuration entails not just the control algorithm but may include changes to pipework or some other physical change.

However, for a particular house, only one configuration can be active and for it to properly take effect, must be active for some period of time (such as a week). Nonetheless, the weather and use of the building may vary from week to week. The problem is how to compare the two configurations A and B?

One solution might be to interleave the configurations on a weekly basis (e.g., ABBA or BAAB) so that during a period whether the weather is changing (from winter to spring, e.g.) then the test is roughly fair to both. A problem, however, is that it may be the case that A performs well only in the current season but B is a better all-rounder. Thus it may be necessary to also test the configurations in a virtual setting and combine this with the real results to make the test fairer.

Another possible solution might be to look at the weather experienced during A versus that during B and to adjust accordingly. This has the difficulty is that it is unclear how to adjust the results to achieve a fair comparison.

Further considerations are how to fairly assess different configurations in terms of energy consumption and thermal environment. For example, is it sufficient to maintain a “comfortable” thermal environment according to some model of human thermal comfort perception, or should a target temperature be achieved.

Possible research questions include:

  1. What is a sufficiently `fair’ way to compare two configurations for a heating / cooling system?
  2. What are the appropriate metrics? For example, is Coefficient of Performance (COP) useful or do we also need to consider energy consumption or energy cost?
  3. How can virtual environments help with ensuring a fair comparison?