LA SENSATEZ COMO TALENTO Introducción
1.3 La sensatez como talento.
1.3.2 Sensatez y talento.
The development of computer models to predict annual energy consumption using hourly simulations and weather data that represents an average year goes back to the 1960s. With the
103 p. 2-328, West, Anne and Mike Logsdon (The Cadmus Group, LLC), Howard Reichmuth, PE, and Jarred
Metoyer (KEMA), “The Field Hand and the California Model: An Example of Blending Operational Measurements with Performance Models,” 2010 ACEEE Summer Study on Energy Efficiency in Buildings.
advent of performance-based codes like Title 24 computer models have become relied on and determine whether a building can be constructed. Simulation is simplified by adopting standardized occupancy, lighting, thermostat and other internal gain scenarios. These can be reasonable but are claimed to be the average. The impact interactions between schedule assumptions and building systems are thought not to impact the calculation of the difference in annual energy use between the proposed and standards building because they are the same except for key components like HVAC efficiency. Accuracy of simulation programs are measured by the degree that they vary from the benchmark program.
Section 6 discusses the difficulties in moving from EER measurements to estimates of annual kWh usage. The analysis of measurement uncertainty presented in Section 7 points to problems with one time measurement accuracy. Even if it were possible to obtain unbiased pre- and post- service EERs the “measured” improvement in EER would not have desired statistical
significance, and more importantly would not provide a solid basis for quantifying the ultimate goal: annual energy and peak demand savings. Even though the algorithms and performance maps have proven suited to the tasks of standardized ratings and practical design of HVAC systems there is no definitive study of uncertainty.
A significant effort was taken by Koran, et al, to calibrate DOE-2.1C to site-monitored data. The authors concluded that they “were able to successfully tune the DOE-2 model to monitored data…so both tuned models estimated annual HVAC energy use within 11% of the monitored consumption.”104 There is no claim made concerning the uncertainty of the model. The paper provides the reader an understanding of the number of difficult steps needed to “tune” the models to come within acceptable ranges. “Inputs are adjusted, within reasonable bounds, until the simulated and monitored HVAC energy uses are within the following limits: each month ±30%; and seasonally ±20%.”105 Other inputs were also adjusted to the point that abnormally high insulation values had to be assumed. It is not a critique of the approach or its
implementation to note that the accuracy of the simulation model cannot be definitively established.
Five increasingly detailed levels of tuned simulations were performed by Alerez and Faramarzi using data from six buildings in the SCE service territory. Level 1 simulations used building- specific data for the inputs to the model. “Absolute estimation errors for HVAC End Use Intensity (EUI) ranged from 1 to 27 percent.”106 The average error was 17.8%. Level 5 reduced the average error in HVAC EUI to 11.6% and “three buildings had reduced absolute errors and three had increased absolute errors”107. In this study the inputs to DOE-2 were not modified to the extent undertaken by Koran. Level 2 used monthly kWh and kW data such as would be available from utility bills. Level 3 modified HVAC schedules and inputs based on highly detailed surveys of the operation and maintenance at each site. Level 4 used whole building hourly energy use data and finally Level 5 used load shapes derived from detailed monitoring of
104 p. 3.175, Koran, W. (PECI), Kaplan, M. (Kaplan Engineering), and Streele, T. (BPA), “DOE-2.1C Model
Calibration with Short-Term Tests versus Calibration with Long-Term Monitored Data,” ACEEE Summer Session, 1992.
105 p. 3.167, Koran 1992.
106 p. 2.14, Alereza, T. (ADM) and Faramarzi, R. (SCE),“ More Data is Better, But How Much is Enough for Impact
Evaluations?” ACEEE Summer Session , 1994. EUI is end-use intensity in kWh/sq.ft.
lighting. It would be of great value to go back to these buildings, re-establish the base line HVAC consumption, make improvements to the HVAC systems and then monitor and simulate the impacts of the improvements.
In parallel with the development in hourly simulation modeling an alternative approach based on kWh, kW, and Therms measured at the utility meter has been developing. Nonintrusive
Appliance Load Monitoring (NALM)108 uses metered data with an inventory of appliances to calculate annual energy use of each appliance. In the early 1990s researchers developed and patented specialized monitoring devices that could be installed without shutting off the building power. Insight into the functionality of NALM is presented in “Non-Intrusive Electrical Load Monitoring, a Technique for Reduced-Cost Load Research and Energy Management”, in which the authors discuss “recent results from a field test of the non-intrusive load monitor (NILM) as applied to the residential sector, followed by a report of ongoing research that focuses on the commercial sector.”109 With the implementation of smart meter technology every building has detailed hourly and sub-hourly data; NALM implementation is possible on every building. Research will be needed to develop how to analytically recognize the energy use of each type of appliance without the super-fine profiles generated by the devices discussed in the 1992 paper. With this done, research can then focus on developing the software needed to output the
breakdown by end use of energy use hourly, daily, monthly and annually at a site. It is then a matter of data analysis to aggregate a large number of sites to derive new level of understanding on HVAC and other end uses. Powerful statistical methods could be used to support energy efficiency program development, implementation and EM&V. As a matter of standard practice customer privacy would be protected and when needed permission would be procured from customers to assess pre and post energy use. Web-based monitoring that is separate from smart meters could add value by giving information to customers that they or their QM contractor can use to save energy and reduce peak demand permanently. It will be worth the efforts of a task force of stakeholders to work out the complex issues of how to make “universal” NALM work to the benefit of customers who will be faced with increasing economic pressure to control peak demand.