Annual Energy Prediction Research Articles

A new workflow for load forecasting in planning areas: Considering uncertainties in building scenarios

Urban Building Energy Model (UBEM) is a widely used method for load forecasting in urban energy planning. It typically depends on reference buildings to simplify particular scenarios and uses energy calculation software for load forecasting. However, significant challenges exist when using this approach in unbuilt areas, because there is no established distribution of buildings or a widely recognized database of reference buildings. To overcome this technical barrier, this paper introduces a new workflow - Scenario Urban Building Energy Model (S-UBEM). This approach utilizes publicly available databases to establish a reference building database and a stochastic human behavior database. By utilizing these databases and planning documents, the methodology can construct building energy use scenarios for the undeveloped area, ultimately achieving load forecasting. In this paper, the proposed method is validated by transforming the built-up area into an unbuilt state. The results demonstrate that S-UBEM more accurately represents the spatial and temporal distribution of energy consumption. This is evidenced by a 28.3 % reduction in the annual energy prediction error compared to the conventional method, as well as by visualizing the distribution of energy consumption at the building level.

Bayesian uncertainty quantification framework for wake model calibration and validation with historical wind farm power data

AbstractThe expected growth in wind energy capacity requires efficient and accurate models for wind farm layout optimization, control, and annual energy predictions. Although analytical wake models are widely used for these applications, several model components must be better understood to improve their accuracy. To this end, we propose a Bayesian uncertainty quantification framework for physics‐guided data‐driven model enhancement. The framework incorporates turbulence‐related aleatoric uncertainty in historical wind farm data, epistemic uncertainty in the empirical parameters, and systematic uncertainty due to unmodeled physics. We apply the framework to the wake expansion parametrization in the Gaussian wake model and employ historical power data of the Westermost Rough Offshore Wind Farm. We find that the framework successfully distinguishes the three sources of uncertainty in the joint posterior distribution of the parameters. On the one hand, the framework allows for wake model calibration by selecting the maximum a posteriori estimators for the empirical parameters. On the other hand, it facilitates model validation by separating the measurement error and the model error distribution. In addition, the model adequacy and the effect of unmodeled physics are assessable via the posterior parameter uncertainty and correlations. Consequently, we believe that the Bayesian uncertainty quantification framework can be used to calibrate and validate existing and upcoming physics‐guided models.

Comparison of platform and sea-bed mounted flow measurement instrumentation for SME PLAT-I

Tidal resource assessment for the characterisation of turbine performance or Annual Energy Prediction currently uses the method of bins as recommended by international standards. An alternative method is proposed in this paper and applied to the Sustainable Marine Energy PLAT-I deployment in Connel Sound, Scotland. This method may be suitable for tidal turbines which operate from the surface. Three instrumentation types are used in this work, a bed-mounted Acoustic Doppler Profiler (ADP), and platform-mounted Acoustic Doppler Velocimeter (ADV) and Electromagnetic Current Meter (ECM). By comparing the resource characteristics from these three sources, a comparison of their velocity magnitudes and turbulence characteristics is made, demonstrating the difference between methodologies. It was found that the ADP evaluated using the method of bins produced a more conservative velocity distribution, in comparison to the ADV and ECM. Consequently, a representative AEP showed a difference of 3.8kWh (50% of ADP total) for the month of data collected. When comparing the Turbulence Intensity between devices, the ADP and ECM had similar metrics whilst the ADV had up to 14% higher values. The significance of these differences requires further work comparing them to the SME PLAT-I turbines power output to ascertain which best represents the onset flow experienced by the turbine and if there is a correlation between power performance and turbulence intensity.

Uncertainty of building energy performance at spatio-temporal scales: A comparison of aggregated and disaggregated behavior models of solar shade control

The impact of occupant uncertainty and inter-occupant diversity of solar shade control on building energy performance at spatio-temporal scales was investigated. A stochastic model of manual solar shades developed in a previous study based on field measurements was used in this paper. Co-simulation, statistical calculation and uncertainty analysis were conducted. Comparison of energy prediction uncertainty between aggregated and disaggregated behavior models was performed to determine whether occupant behavior and related energy prediction were suppressed by the aggregated behavior model. The results show that predicted energy uncertainty due to occupant behavior varies largely at different spatio-temporal scales ranging from 0.8% to 160%. Occupants’ impact should be considered at small spatio-temporal scales due to high energy uncertainty (can reach more than 100%) while this effect can be ignored at large spatio-temporal scales (the uncertainty is less than 2% at annual scale). The suppressing effect by the aggregated model on shade behavior uncertainty is significant, but it is not significant on resulting energy uncertainty, especially for large buildings and annual energy prediction. Aggregated model for the developed shade behavior model can be used instead of disaggregated model in the prediction of annual energy demand of large buildings.

A combined model for PV system lifetime energy prediction and annual energy assessment

This paper presents a generic model for the prediction of the lifetime energy production of photovoltaic (PV) systems and the assessment of their annual energy yield in different time periods of operation. As case studies, it considers domestic PV system generation potentials in the UK and India to demonstrate the model results across a range of contrasting climatic and operating conditions. The model combines long-term averages of solar data, a commercial PV system simulation package and a probability density function to express the range of the annual energy prediction in different time periods of system operation. Moreover, a sensitivity analysis based on degradation rates and energy output uncertainties is embedded in the lifetime energy calculations. The importance of the reliability and maintenance of the PV systems and the energy prediction risks, especially regarding economic viability, are demonstrated through the PV lifetime energy potentials in these two countries. It is shown that, even for countries that are significantly different in respect to their solar resource, PV systems may produce similar amounts of energy during their lifetime for reasonable assumptions of degradation rates and uncertainty levels.

ゼロ・エネルギー・ハウスの実現に向けた実証研究 （第2報）対象住宅の断熱性能・室内環境実測および年間エネルギー予測

ゼロ・エネルギー・ハウスの実現に向けた実証研究 （第2報）対象住宅の断熱性能・室内環境実測および年間エネルギー予測

Numerical and experimental results of a novel and generic methodology for energy performance evaluation of thermal systems using renewable energies

At present there is no reliable approach to model and characterize thermal systems using renewable energy for building applications based on experimental data. The results of the existing approaches are valid only for specific conditions (climate type and thermal building properties). The aim of this paper is to present a generic methodology to evaluate the energy performance of such systems. Artificial neural networks (ANNs) have proved to be suitable to tackle such complex problems, particularly when the system to be modelled is compact and cannot be divided up during the testing stage. Reliable “black box” ANN modelling is able to identify global models of the whole system without any advanced knowledge of its internal operating principles. The knowledge of the system’s global inputs and outputs is sufficient. The proposed methodology is applied to evaluate three different Solar Combisystems (SCSs) combined with a gas boiler or a heat pump (HP) as an auxiliary system. The results show that the best ANN models were able to predict with a satisfactory degree of precision, the annual energy consumption of the all systems except the SCS combined with air source HP, in different conditions, based on a learning sequence lasting only 12days. In fact, the annual energy prediction errors were less than 10% in most cases. The methodology limitations appear in extreme boundary conditions (Barcelona climate) compared to those used during the ANN training process.

Changes in Cadmium Telluride Photovoltaic System Performance due to Spectrum

Seasonal and short-term weather-related changes in the solar spectrum can induce shifts in the performance of photovoltaic (PV) systems that affect both annual energy predictions and system characterization. The spectral shift factor, which is a metric indicative of how much the performance of a PV system will vary from nameplate due to deviations from the ASTM G173 spectrum (air mass of 1.5), is predicted using TMY3 data and the simple model of the atmospheric radiative transfer for sunshine (SMARTS) model and is correlated with cadmium telluride (CdTe) PV system performance in four different climates. The predicted spectral shift factors for CdTe systems show improved performance in the late summer and early fall and diminished performance in the winter. These intraannual variations can be as large as ±3%, but annual spectral shift factors are typically within ± 1% of nameplate. The spectral shift factor of CdTe systems was found to be most sensitive to the precipitable water content of the atmosphere. Consequently, a parameterization of CdTe spectral shift factor as an exponential function of precipitable water is derived using the outputs of the SMARTS model in 11 locations. This parameterization is shown to predict observed monthly and daily fluctuations in CdTe PV performance. Future efforts will incorporate this methodology into energy predictions that will reduce uncertainty.

Intelligent Tools for Wind Resource Correlation

The use of linear correlation techniques to extend monitoring mast records is an important step in developing the energy estimates that underpin wind development business cases. Yet accuracy of the linear correlation technique rapidly degrades in complex conditions, and only allows correlation of a small number of parameters without introducing overwhelming model complexity. This paper argues that artificial neural network (ANN) models offer an alternative that is simple to use, afford greater correlation options and provide improved data extension accuracy. This was explored through a practical data extension example, with the linear and ANN correlation performances directly compared. The case study showed that the ANN correlation method provided superior accuracy compared to the linear correlation method, including a three fold improvement in Annual Energy Prediction (AEP) error.

Comparison of Predicted to Measured Photovoltaic Module Performance

To accurately predict the electrical performance of photovoltaic modules computer simulation models are essential. Without such models, potential purchasers of photovoltaic systems have insufficient information to judge the relative merits and cost effectiveness of photovoltaic systems. The purpose of this paper is to compare the predictions of a simulation model, developed by Sandia National Laboratories, to measurements from photovoltaic modules installed in a vertical wall façade in Gaithersburg, MD. The photovoltaic modules were fabricated using monocrystalline, polycrystalline, tandem-junction amorphous, and copper-indium diselenide cells. Polycrystalline modules were constructed using three different glazing materials: 6 mm low-iron glass, 0.05 mm ethylene-tetrafluoroethylene copolymer, and 0.05 mm polyvinylidene fluoride. In order to only assess the simulation model’s ability to predict photovoltaic module performance, measured solar radiation data in the plane of the modules is initially used. Additional comparisons are made using horizontal radiation measurements. The ability of the model to accurately predict the temperature of the photovoltaic cells is investigated by comparing predicted energy production using measured versus predicted photovoltaic cell temperatures. The model was able to predict the measured annual energy production of the photovoltaic modules, with the exception of the tandem-junction amorphous modules, to within 6% using vertical irradiance measurements. The model overpredicted the annual energy production by approximately 14% for the tandem-junction amorphous panels. Using measured horizontal irradiance as input to the simulation model, the agreement between measured and predicted annual energy predictions varied between 1% and 8%, again with the exception of the tandem-junction amorphous silicon modules. The large difference between measured and predicted results for the tandem-junction modules is attributed to performance degradation. Power measurements of the tandem-junction amorphous modules at standard reporting conditions prior to and after exposure revealed a 12% decline. Supplying post exposure module parameters to the model resulted in energy predictions within 5% of measured values.

Numerical procedure for predicting annual energy consumption of the under-floor air distribution system

As compared with the mixing system, indoor air temperature stratification in the under-floor air distribution (UFAD) system offers an opportunity for cooling load reduction in the occupied zone. This stratification is a major feature that offers the energy saving potential, but it has not been thoroughly taken into account in most energy simulation programs. In this article, a numerical procedure, based on coupling two types of modeling, i.e., CFD (computational fluid dynamic) simulation and dynamic cooling load simulation, is proposed to predict annual energy consumption. The dimensionless temperature coefficient is first defined in the UFAD system and obtained from CFD simulation, based on the boundary conditions from a cooling load program ACCURACY. According to this coefficient, temperature stratification input to ACCURACY is then revised to calculate the updated supply and exhaust air temperatures for final annual energy prediction. To demonstrate the method, a small office room is investigated using Hong Kong weather data. With the constant air volume (CAV) supply in the UFAD system, it is found that the dimensionless temperature coefficient is almost a constant, when the locations of heat sources are fixed. As compared with the mixing system, the UFAD system derives its energy saving potential from the following three factors: an extended free cooling time, a reduced ventilation load, and increased coefficients of performance (COP) for chillers.

The early design model for prediction of energy and cost performance of building design options

In order to encourage the use of computer modelling in building environmental analysis, it is necessary to provide a model developed from the designer's point of view. Detailed simulation models require a high degree of expertise and familiarity, further, there is also a need for detailed information not available in the early stages of the design process. Simplified models play an important role in the early stages of a design to achieve an integrated design: firstly, they are easy to use and, secondly, they require information easily available at the start of a design. In the Early Design Model (EDM) the solar gain utilisation factor has been determined as a continuous function of thermal mass. The differences between the annual energy predictions of EDM and SERI-RES ranges from 0.1% to 4.6% for time constants ranging from 378 to 2.52 hours. The differences between the two sets of predictions on monthly basis ranges from m 3.6% to m 6.48% (EDM's predictions being larger) during the heating season, and from +2.86% to a maximum of +51% (EDM's predictions being smaller) in the remaining part of the year. In addition to energy predictions, EDM incorporates a facility which gives cost indications.

THE EARLY DESIGN MODEL FOR PREDICTION OF ENERGY AND COST PERFORMANCE OF BUILDING DESIGN OPTIONS

In order to encourage the use of computer modelling in building environmental analysis, it is necessary to provide a model developed from the designer's point of view. Detailed simulation models require a high degree of expertise and familiarity, further, there is also a need for detailed information not available in the early stages of the design process. Simplified models play an important role in the early stages of a design to achieve an integrated design: firstly, they are easy to use and, secondly, they require information easily available at the start of a design. In the Early Design Model (EDM) the solar gain utilisation factor has been determined as a continuous function of thermal mass. The differences between the annual energy predictions of EDM and SERI-RES ranges from 0.1% to 4.6% for time constants ranging from 378 to 2.52 h. The differences between the two sets of predictions on monthly basis ranges from −3.6% to −6.48% (EDM's predictions being larger) during the heating season, and from +2.86% to a maximum of +51% (EDM's predictions being smaller) in the remaining part of the year. In addition to energy predictions, EDM incorporates a facility which gives cost indications.

Energy load predictions for buildings based on a total demand perspective

The outline of this work was to develop models for single family buildings, based on a total energy demand perspective, i.e., building-climate-inhabitants. The building-climate part was included by using a commercial dynamic energy simulation software. Whereas the influence from the inhabitants was implemented in terms of a predicted load for domestic equipment and hot water preparation, based on a reference building. The estimations were processed with neural network techniques. All models were based on access to measured diurnal data from a limited time period, ranging from 10 to 35 days. The annual energy predictions were found to be improved, compared to models based on only a building-climate perspective, when the domestic load was included. For periods with a small heating demand, i.e., May-September, the average accuracy was 7% and 4% for the heating and total energy load, respectively, whereas for the rest of the year the accuracy was on average 3% for both heating and total energy load.