Part Three: The measurable effects of unconsidered user behavior

From: Smart office buildings – curse or blessing? Human needs and “smartification”

How can robustness analysis be used to estimate the impact of incorrect assumptions in building design processes? What empirical data is there? In contrast to the first parts of this series of articles “Smart office buildings – curse or blessing?”, the focus in this report is not on human needs and behavior, but rather on their impact on energy consumption.

Newly built and renovated buildings today are highly optimized in terms of energy efficiency, at least as far as economics allow. A model is used to forecast the energy consumption of a planned building. Models are always more or less detailed representations of reality. Assumptions often have to be made, for example about future usage intensity. This is necessary and correct, because otherwise no statements at all would be possible about a building that has not yet been completed.

However, deviations from the assumptions or input parameters of the calculation do not lead to a proportional, but to a non-linear change in energy consumption. This is interesting because buildings, depending on their construction and technical system, react differently to changes in input parameters. The following studies show that highly technical and efficient buildings tend to react very sensitively; small changes in the input parameters lead to a large increase in energy consumption.

Robustness analysis provides a method for evaluating the impact of assumptions made or their possible deviations in reality. The following paragraphs first explain the context and then show how this method can be used to estimate the impact of incorrect assumptions about human behavior in buildings.

For the first, general assessment of the effects, we do not need any complex analyses, but can rely on statistical data about measured energy consumption and the energy requirements previously calculated in the planning.

The consumption/demand variance

First of all, unconsidered human behavior leads to a consumption/need deviation due to rebound effects. Empirical surveys show that the measured energy consumption in highly efficient buildings is approx. 5% above the energy requirement forecast for this building according to EnEV (Energy Saving Ordinance). This means that more energy is consumed than previously calculated. In contrast, the average consumption of less efficient buildings is more than 25% below the predetermined values. This is shown in the following figure, which shows the consumption factor (calculated as measured consumption divided by the calculated demand) for the different efficiency classes of the energy certificate (Insulation of the German building stock). In particular, people reduce the effect of technology through their behavior - in better insulated buildings, for example, heating is more unrestrained because it is less expensive (or less harmful to the climate) than in an uninsulated old building (rebound effect).

In addition to the interior temperature, hot water use, ventilation behavior, night setback and partial heating were also identified as further influencing factors through which human behavior influences real energy consumption.

Let us now turn our attention to the robustness analysis, with which the TU Munich takes a different approach to assess the effects of fluctuations in the input factors - such as human behavior.

Robustness in construction

The term robustness is used in many disciplines - described, for example, as "error tolerance" (computer science), "on-time production despite undesirable influencing factors" (production), "reproducible and standardizable results despite variability in the sample to be analyzed" (diagnostics) or even "evolutionary stability of a certain characteristic” in biology.

In its research series “Simply Building”, the TU Munich uses the term “robust optimization”, in which “a system in the sense of a design or process (technology, product) is described as robust if it can withstand the fluctuations of the Input factors are minimally sensitive or insensitive.” And explains: “The graph [see cover photo, note d. Author's] shows the difference between a global and a robust optimum - reduced to two parameters: the uncertain input variable (Δx) and the target variable of the function influenced by it (Δy). The result of the global optimum (Δy1) is significantly influenced by the fluctuation of the input variable. In contrast, the same spread of the input variable (Δx) for the robust optimum (x2) only has a minor effect on the result (Δy2)." (Easy to build 1)

Uncertain boundary conditions in robustness analysis

The Technical University of Munich initially determined the energy consumption for individual rooms in buildings of different constructions for a “best case”. The six construction variants considered are:

• Four easy-to-build variants:
  • masonry material,
  • Lightweight concrete made from one material,
  • Wood hybrid, and
  • Solid wood (with sun protection),
• a standard variant according to EnEV 2014/16 (“Standard”), as well as
• a low-energy/passive house standard variant (“low energy”).

Then, contrary to the usual (normatively correct) assumption, the following four input parameters were classified as “uncertain” and changed “extremely” for the energy consumption determination:

• the climate,
• user behavior,
• the internal profits or burdens, as well
• the sun protection. 

Even if the Technical University of Munich speaks of “extreme” changes here, the scenarios seem quite realistic: milder weather due to climate change by 2045, people do not ventilate permanently, internal loads (e.g. laptops and screens) doubled to around 20W/m², and failure of the sun protection.

With regard to human behavior, the criteria “people do not ventilate / permanently”, but also “failure of the sun protection” are interesting, since “failure” is roughly equivalent to “overcontrolled”. 

While – as expected – the heat required increases through continuous ventilation in all construction variants, the effects are highest in a “low-energy house”. These variants are equipped with a mechanical ventilation system and are therefore the most efficient in the normative calculation. However, an open window can disrupt this concept so much that the energy consumption in this scenario exceeds all other variants. (See Figure 2) (Easy to build 1)

In addition to energy consumption, the robustness analysis at the Technical University of Munich also analyzed comfort with regard to the indoor temperature in summer (“over-degree hours”) (see Fig. 3). This is where additional internal loads (e.g. in the form of electrical devices such as laptops and screens) have the greatest impact. High-tech buildings also react most sensitively to this, as well as to the failure of the sun protection. (Easy to build 1)

Application of robustness analyzes in the building planning process

Robustness analyzes can make a valuable contribution to future-proof building planning and design. As part of thermodynamic building simulations, which are common for larger multi-family and non-residential buildings, many of the criteria considered here can be easily varied. If well prepared, the results enrich decision-making templates for developers and can thus lead to increased awareness of the topic.

Conclusion

Assumptions must be made in the building design process. Many of these assumptions have a decisive impact on the basic design and selected technical equipment. With the robustness analysis, we come one step closer to the goal of being able to safely maintain energy efficiency and comfort in buildings at a high level. At the same time, we can evaluate our assumptions and their possible deviations in terms of their impact and, if necessary, correct them.    

The impact of unaccounted human behavior on the energy consumption of buildings is statistically approximately +5%/-25%. The Technical University of Munich shows that “extreme” behavior can lead to a doubling or halving of energy consumption, depending on the type, and that high-tech buildings react most sensitively. A similar picture emerges when considering summer thermal protection, one of the main criteria for comfort.

If one wants to compare the results of this empirical determination with those of the robustness analysis in detail, some inconsistent factors or criteria become apparent. In the robustness analysis by the Technical University of Munich, interior temperature, hot water use, night setback and partial heating are missing, while ifeu/Beuth did not consider internal loads and the failure or overcontrol of the sun protection. Completing these criteria could be part of further research.