OBJECTIVE TEUI3 and Occam’s Razor

Embracing Parsimony: The OBJECTIVE TEUI 3 Calculator’s Simplicity in Architectural Energy Models

Occam’s Razor, the principle often attributed to the medieval philosopher William of Ockham, advocates for simplicity: where the simplest solution is usually the correct one. This philosophical approach has profound implications across various fields, from physics to economics, and architecture and energy modeling are a mix of both of these disciplines! OpenBuilding’s’ (OAA) TEUI 3 calculator epitomizes this principle of parsimony or economy, by effectively cutting through the complexity of traditional dynamic, hourly energy models to provide clear and accurate assessments of building performance, by starting with Actual Total Energy Use Intensity read from Utility Bills. With a single page of inputs, and with just a few dozen user inputs (we are working on automating export from BIM models), OBJECTIVE is easy to complete and even easier to review.

The Principle of Parsimony

Occam’s Razor is not a modern invention. The concept can be traced back to Aristotle, who said “We may assume the superiority [other things being equal] of the demonstration which derives from fewer postulates or hypotheses.” (https://en.wikipedia.org/wiki/Occam’s_razor) Ptolemy, the Greco-Roman mathematician and astronomer, also emphasized simplicity in his works, striving for models that explained celestial movements with minimal assumptions, “We consider it a good principle to explain the phenomena by the simplest hypothesis possible.”

In the context of energy modeling, the principle of economy guides the OBJECTIVE calculator. By starting with actual results of building performance through a review of utility bills, the TEUI 3 avoids the pitfalls of models and performance standards that rely only on projections of performance.

We have often stated that Olympic Medals are based on Actual, and never Simulated performance, so why should green buildings be any different? This approach aligns with a broader understanding that complexity does not necessarily equate to accuracy and even a good guess is no match for truth.

One need not measure the fart of a passing mouse to add to interior gains to determine an accurate picture of gains and losses. Similarly we don’t need to know what the weather is doing at every hour of the year to know what the annual HDD and CDD will have on the totals. Some things are trivial, where others are not. Some things add complexity with no net effect on accuracy. Understanding these differences is key to solving the performance gap, or the discrepancy between Modelled performance and Actual performance. With the actual TEUI obtained from energy bills (TEUI = Energy/Area), we can then separate out probable (or where these are normalized by the Building Code) SHW, Plug, Light and Equipment loads and see whether the thermal loads correlate to a model of linear regression (for more on that see: https://www.energylens.com/articles/degree-days).

Thermal Loads are often driven by Heating and Cooling Degree Days (HDD, CDD). Thousands of energy models show that the principal driver of total loads, once the daily, baseline (constant) loads are isolated, is climate. This is why it is absolutely vital to use current, recent past, and projected/future data sets for climate references. In many cases the data cited in Building Codes and Standards can date as far back as 1951, or be based on averages from 30 years ago. We have current weather data, annualized up to the previous year, and we have really good projections for future weather based on current climate science and recent precedents for extreme scenarios – so shouldn’t we be using this data? Shouldn’t we be at the very least, looking at it?

The magnitude of the load correlates to the magnitude of the loss, as a function of airtightness, thermal resistance, and the gains in the cooling or heating season. This is known as the Energy Balance, what goes in, must come out – such is universal equilibrium.

Wait, what?

Let’s say you have a cooler of beer and nice fresh, ice. On a Summer day, with 30ºC and 50% RH, you might want to know how long will this R10 cooler of a certain size, with a certain volume, and a certain number of cans and kg of ice, stay cold – or reach its interior equilibrium temperature and then drop? When will the ice melt? This is the fundamental mechanism inside the TEUI3 tool – the determination of the rate of losses based on exterior and interior temperatures. But we should never turn our brains off. Intuition and experience tell us the ice will typically melt in less than 24 hours, and our beer will remain cold really only shortly after the last ice cube melts. So if the calculations tell us the beer will be cold in a week, or it will be warm in 3 hours – we know we’re off track.

Obviously the whole situation changes when Fall arrives, and the outdoor temperature is only 10ºC – the beer stays colder longer because the Delta-T (inside and outside temperature difference) is smaller. Maybe I want to switch to wine. Forget the ice. Put a sweater on. We respond to situations, and buildings should too. So how can we do a better job at making buildings responsive? We can do this with TEUI3 because it is easy to test scenarios with many different ‘switches’ and features, from ventilation rates, to shading factors, to equipment efficiencies, to fuel-switching, and the results are immediately apparent in the header line. No need to run analysis and then wade through dozens of pages of reports to review the totals.

The Complexity Conundrum

Traditional energy models involve intricate calculations and often too many assumptions about building performance – especially when it comes to user behaviour, schedules, pumps, fans, and pipe losses, etc. One very common scenario is that the modeller, charged with creating a reference model, does not alter all of the assemblies to reflect actual designed R-values or airtightness targets or occupant activity or durations/schedules, but relies on code minima and makes a few small tweaks to hit some target, then calls it a day. This can result in a significant over-sizing of mechanical equipment due to an over-estimation of loads.

Or a modeller can use some default value for the fraction of useful gains (sometimes known as an n-factor for gains, this happens automatically in PHPP), without questioning whether such a factor is relevant for the occupancy type. Take a movie theatre for example, you have a sudden influx of an audience, hundreds of people contributing up to 100 Watts each, and exhaling CO2, perspiring and all the rest. That sudden gain is not useful from mindnight to noon the next day, but only during the occupied period – so the usefulness of these gains are relatively low, less than the occupied hours in a day, say 9/24 = 37.5%, and since not all of that gain can get captured for heating during the occupied period, the useful gain actually trends towards 0%.

Similarly, a huge influx of solar gains on a sunny October afternoon cannot be used to heat a building on a January morning – those gains are time-dependent and very often exhausted from an open window as ‘unwanted gains‘ at that time, but this is what happens in model-land when more than 50% of heating season gains are considered useful.

Or a modeller can use the wrong climate data, or outdated weather data, or weather data that doesn’t consider extremes or even future weather scenarios, or weather data using the wrong HDD or CDD baselines – which can throw a model’s target value estimation up to 30% or more off of actual utility bill values.

Energy models can become convoluted, with each added layer of complexity increasing the potential for errors – especially when errors result from a human being entering and reviewing the data for analysis, without careful deliberation of what the data means, or what the math under the hood is doing. This has pushed energy modelling into whole new discipline of expertise that has divorced architects and other designers of buildings from performance modelling altogether. As programmers often joke, “When I first wrote the code, only God and I knew how it worked, but now, only God knows.” Not knowing how so many ‘black-box’ energy modelling platforms actually work can lead to a kind of automation complacency, and the result is the well-known performance gap between modelled and actual performance.

TEUI3 aims to re-introduce performance modelling, and the fundamentals of building physics to the designer for early stage design, where some of the most significant improvements with respect to massing, volumes, surfaces, orientation, shading, overall R-values and equipment efficiencies can be established so that targets can be meaningfully established.

The pressure on an external ‘modelling expert’ to shape data to fit a target can be very high, especially when a project is targeting a client-driven performance metric but the rest of the design team has not participated in any form of performance improvement exercise. At this point, the handoff to a modelling professional can best be described as the handover of a ’turd awaiting polishing’. 

The financial bias or the inherent conflict of interest of modelling professionals tasked with meeting performance goals when the design team has not been tasked with optimizing performance, has led some to propose a code of ethics for individuals working in the simulation space. (https://www.site.uottawa.ca/~oren/pubs/pubs-2002-03-Code.pdf). Architects on the other hand, already operate with the public interest as one of their principal aims, and also operate under a code of ethics.

In economics, “A study of the predictive validity of Occam’s razor found 32 published papers that included 97 comparisons of economic forecasts from simple and complex forecasting methods. None of the papers provided a balance of evidence that complexity of method improved forecast accuracy. In the 25 papers with quantitative comparisons, complexity increased forecast errors by an average of 27 percent.” https://en.wikipedia.org/wiki/Occam’s_razor

This finding underscores a critical point: complexity does not inherently improve the accuracy of predictions. In fact, it often does the opposite. By embracing simplicity,  TEUI 3 leverages real-world data to construct a clear picture of energy losses, gains, and emissions based on actual purchased energy, and only then attempts to parse these into their most likely categories, based on empirical evidence of losses and gains, and a limited number of simple, physical/mathematical projections, to derive the energy balance for both cooling and heating seasons. 

One example; when Service Hot Water (SHW) is 100% gas-fired, and the EF (efficiency) of the equipment is known, and all other loads are electric, isolating SHW loads becomes easier – it’s the ekWh of the gas consumed divided by the number of occupants. A building with double the occupants will very likely have double the gas usage, or electrical usage as the case may be (which is why TEUI3 converts all loads to ekWh – to share a common unit). This progressive isolation of known values from unknown values across dozens of buildings helps to refine TEUI3’s targeting fidelity.

Practical Applications

The TEUI 3 calculator’s method starts with the straightforward step of reviewing utility bills, for electricity, water, gas, oil, propane, biofuels and any renewable contributions such as building-integrated PV or even WWS-based RECs. This initial, backwards-looking review of empirical data eliminates the guesswork involved in blindly estimating energy consumption. By centring performance on actual energy usage, the OBJECTIVE calculator identifies patterns and trends that might be obscured by more complex models.

But then in the forward-looking design process, the next model one builds can be based on established typologies for say, a 3-Storey MURB, using the same systems and envelope metrics. Results are likely to be very similar when OBJECTIVE is in Target mode even before utility bills are known. The training wheels can then come off, but we later validate the Targeted Use the moment we have access to the Utility Bills, and the cycle renews. Each new model is refined by the last, and the performance gap effectively closes. This is why it is important at the outset of a project to require as a contract term that the Architect have access to utility bill data for the purpose of performance verification and quality assurance.

Once this data is established, the calculator then parses out a detailed picture of probable energy losses and gains. This approach ensures that all models remains rooted in reality, providing architects and engineers with reliable information to make informed decisions. The simplicity of the TEUI 3 calculator allows for easier verification, validation, review and ultimately improvements, enhancing the overall trust in the model’s accuracy on past and future buildings.

Reducing Emissions Through Accurate Modeling

An accurate understanding of energy performance is crucial for reducing energy and emissions. The OBJECTIVE calculator’s parsimonious approach helps in pinpointing inefficiencies and areas for improvement. With a focus on actual data and established benchmarks, rather than purely theoretical projections, it offers a practical pathway to reducing energy consumption and, consequently, emissions.

In an era where sustainability is paramount, the OBJECTIVE calculator provides a valuable tool for architects and engineers committed to designing energy-efficient buildings. With Occam’s Razor as a guiding design principle of the tool, this ensures that the focus remains on the fewest possible user inputs, and provides actionable insights derived from real-world performance, cutting through the noise and potential errors of overly complex models. OBJECTIVE is open source, free to use, and has been and continues to be extensively peer-reviewed. As a white-boxed tool, anyone can review the math and physics used, and comments and suggestions for improvement are always welcome. 

Conclusion

OpenBuilding’s OBJECTIVE TEUI3 calculator exemplifies the power of simplicity in energy modeling. By adhering to the principle of parsimony, and complimenting projections always with reviewed performance, it offers a reliable and accurate method for assessing building performance. This approach, rooted in the wisdom of thinkers like Aristotle and Ptolemy, and supported by empirical evidence, demonstrates that sometimes, less truly is more. Embracing this simplicity can empower architects and design teams with a powerful tool that can lead to better early-stage decisions, improved energy efficiency, reduced emissions, improved building economics and a more sustainable future.

Want to learn more? Click the button!