Although contemporary and complex building energy models such as EnergyPlus are highly advanced and predictive, a simplified building energy model has many advantages and applications. Chief among the advantages are adaptability to the measured response of a building and conciseness in computer implementation. Important applications will include automated fault detection. One of the most important aspects of a successful building energy model is the ability to accurately account for the heat capacity of the structure, especially the envelope. As part of the development of a simplified physics-based building energy model, a variety of roof and wall models were considered. Three different generic models were considered for opaque walls and roofs (1) an analytical model with closed-form solution, (2) a time-series solution using the ASHRAE transfer function method, and (3) a lumped capacity model with various numbers of capacitive nodes. This paper will describe an investigation of various simplified models for the transient conduction through building walls. Specifically, how different kinds of wall constructions and materials can change when the conduction load passes through the wall and when it impacts the HVAC system. Understanding how the physics of dynamic envelope loading can change how much energy a building uses for heating or cooling and when room conditioning loads occur is vital for energy savings and proper modeling.

Residential energy applications have become an important domain of cyber-physical systems. These applications provide significant opportunities for end-users to reduce their electricity costs and for the utilities to balance their supply and demand in the most effective way. One of the most important applications is predicting the total energy usage of a house. However, an accurate time-series prediction may require significant amount of data, e.g. per appliance energy consumption values, that need costly installations, data storage units, and computation and communication devices. In this paper, we propose a framework that uses a forward-selection-based input filtering mechanism for residential prediction applications. Our framework can effectively reduce the amount of data required for residential energy prediction without sacrificing prediction performance. We demonstrate that 94% of the houses can leverage our method, which leads to up to 80% reduction in required data, greatly reducing the system cost and overhead.

The systematic evaluation of wind turbine drivetrains using hardware-in-the-loop strategies (previously presented in Part I) is demonstrated using a state-of-the-art multi-MW drivetrain design and a 7.5-MW test bench. The test bench has the capability to apply both torque and non-torque loads to the electro-mechanical drivetrain. The proposed method to evaluate the capability of a test bench to impose the loads of interest uses design loads of the drivetrain and the test bench load application unit limits as inputs. The design loads are defined by stochastic time series of the longitudinal, lateral, and vertical forces as well as the yawing and nodding bending moments that the load application unit can concurrently apply to the drivetrain (i.e., combined loading). A total of 14 time series sets are considered to capture the minimum and maximum values of the longitudinal, lateral, vertical, and resultant forces as well as the yawing, nodding, and resultant moments. These time series are processed individually to calculate two metrics: the coverage and the capability ratio of the test bench. The former is a percentage of the time series that be applied by the test bench, and the latter indicates an excess (or deficit) in load application capability as compared with the selected design loads. The results are presented and interpreted using the previously described methodology. The findings suggest a good match between test bench capability and the loads of interest in general, and also points to challenges. These discoveries establish a basis for the experimental verification and the development of compensation methods to enhance test bench capabilities.

The performance of CAES is evaluated for various configurations, with and without thermal energy storage. First, a conventional compressed air energy storage process is modeled using a time series iterative forward differencing method to simulate the round trip efficiency, exergy storage, cavern temperatures and pressures, and the gas expander exit temperature of a CAES plant. The computational model was validated experimentally by comparing trended data of the compression cycle of a 280 HP Gardener-Denver tandem horizontal two-stage compressor to computational results. It was found that the process of cooling the compressors resulted in a large exergy loss and the inefficiencies of the expanders lead to higher temperature gas being exhausted back to ambient pressures. Second, Advanced Adiabatic Compressed Air Energy Storage (AACAES) was simulated to study the effectiveness of storing the thermal energy removed from the compressors to be added to the compressed air as it enters the expanders at a later time. Third, the concept of increasing the capacity of the thermal energy storage systems to allow recharge with concentrated solar heat was explored. It was found that the thermal efficiency of converting the solar thermal energy to power would be high (> 60%). Further, the expander exhaust temperature and exergy are high (> 500 K), implying that additional waste heat energy recovery will be possible. Taken together, the results of this study show that an integrated, high efficiency, on-demand, water-free, solar energy delivery system is possible if combined with an AACAES system.

The performance of dry-cooled ORC systems varies significantly with environmental conditions, diurnally and seasonally. This study analyzes the impact of environmental conditions on overall performance and operational control of dry-cooled ORC systems with fixed and variable geometry expanders. Performance analysis results are extended to time-series environmental data to determine the expected performance of a system over long durations. Analysis results indicate that, if other factors are held constant, variable speed condenser fans produce minimal performance improvement, while variable-geometry expanders have a significant impact.

The cost reduction potential of solar power towers (SPT) is an important issue concerning its market introduction. Raising the steam process temperature and pressure can lead to a cost reduction due to increased overall plant efficiency. Thus, for new receiver configurations a supercritical steam cycle operated at 300 bar / 600°C / 610°C live steam conditions was assumed. The considered systems include innovative direct absorption receivers, either with conventional or beam down heliostat field layouts. For the beam down option the receiver is assumed to be a cylindrical vessel with a flow-through porous absorber structure at the internal lateral area of the cylinder. The direct absorption receiver option consists of a cylindrical barrel with downwards oriented aperture, whose absorber structure at the internal lateral area is cooled by a molten salt film. For the assessment, CFD based methods were developed to be able to examine the receiver efficiency characteristics. Based on the receiver thermal efficiency characteristics and the solar field characteristics the annual performance is evaluated using hourly time series. The assessment methodology is based on the European Concentrated Solar Thermal Roadmap (ECOSTAR) study and enables the prediction of the annual performance and the levelized cost of electricity (LCOE). Applying appropriate cost assumptions from literature the LCOE were estimated for each considered SPT concept and compared to tubular receiver concepts with molten salt and liquid metal cooling. The power level of the compared concepts and the reference case is 200 MW el . The sensitivity of the specific cost assumptions was analyzed. No detailed evaluation was done for the thermal storage, but comparable storage utilization and costs were assumed for all cases. At optimized plant parameters the results indicate a LCOE reduction potential of up to 0.5% for beam down and of up to 7.2% for the direct absorption receiver compared to today’s state of the art molten salt solar tower technology.

High penetration of variable wind and solar electricity generation will require modifications to the electric power system. This work examines the impacts of variable generation, including uncertainty, ramp rate, ramp range, and potentially excess generation. Time-series simulations were performed in the Texas (ERCOT) grid where different mixes of wind, solar photovoltaic and concentrating solar power provide up to 80% of the electric demand. Different enabling technologies were examined, including conventional generator flexibility, demand response, load shifting, and energy storage. A variety of combinations of these technologies enabled low levels of surplus or curtailed wind and solar generation depending on the desired penetration of renewable sources. At lower levels of penetration (up to about 30% on an energy basis) increasing flexible generation, combined with demand response may be sufficient to accommodate variability and uncertainty. Introduction of load-shifting through real-time pricing or other market mechanisms further increases the penetration of variable generation. The limited time coincidence of wind and solar generation presents increasing challenges as these sources provide greater than 50% of total demand. System flexibility must be increased to the point of virtually eliminating must-run baseload generators during periods of high wind and solar generation. Energy storage also becomes increasingly important as lower cost flexibility options are exhausted. The study examines three classes of energy storage — electricity storage, including batteries and pumped hydro, hybrid storage (compressed-air energy storage), and thermal energy storage. Ignoring long-distance transmission options, a combination of load shifting and storage equal to about 12 hours of average demand may keep renewable energy curtailment below 10% in the simulated system.

This work presents an alternative, time-window invariant metric for evaluating the quality of solar forecasting models. Conventional approaches use statistical quantities such as the root-mean-square-error and/or the correlation coefficients to evaluate model quality. The straightforward use of statistical quantities to assign forecasting quality can be misleading because these metrics do not convey a measure of the variability of the time-series included in the solar irradiance data. In contrast, the quality metric proposed here, which is defined as the ratio of solar uncertainty to solar variability, compares forecasting error with solar variability directly. By making the forecasting error to variability comparisons for different time windows, we show that this ratio is essentially a statistical invariant for each forecasting model employed, i. e., the ratio is preserved for widely different time horizons.

Reliable short-term predictions of the wind power production are critical for both wind farm operations and power system management, where the time scales can vary in the order of several seconds, minutes, hours and days. This comprehensive study mainly aims to quantitatively evaluate and compare the performances of different Box & Jenkins models and backpropagation (BP) neural networks in forecasting the wind power production one-hour ahead. The data employed is the hourly power outputs of an N.E.G. Micon 900-kilowatt wind turbine, which is installed to the east of Valley City, North Dakota. For each type of Box & Jenkins models tested, the model parameters are estimated to determine the corresponding optimal models. For BP network models, different input layer sizes, hidden layer sizes, and learning rates are examined. The evaluation metrics are mean absolute error and root mean squared error. Besides, the persistence model is also employed for purpose of comparison. The results show that in general both best performing Box & Jenkins and BP models can provide better forecasts than the persistence model, while the difference between the Box & Jenkins and BP models is actually insignificant.

Wind power forecasting will play a more important role in electrical system planning with the greater wind penetrations of the coming decades. Wind will most likely comprise a larger percentage of the generation mix, and as a result forecasting errors may have more significant effects on balancing operations. The natural uncertainties associated with wind along with limitations in numerical weather prediction (NWP) models lead to these forecasting errors, which play a considerable role in the impacts and costs of utility-scale wind integration. The premise of this project was to examine errors between the actual and commercially forecasted power production data from a typical wind power plant in the Northwestern United States. An exhaustive statistical characterization of the forecast behavior and error trends was undertaken, which allowed the most important metrics for describing wind power forecast errors to be identified. This paper presents only the metrics considered by the authors to be most significant. While basic information about wind forecast accuracy such as the mean absolute error (MAE) is valuable, a more detailed description is useful for system operators or in wind integration studies. System planners have expressed major concern in the area of forecast performance during large wind ramping events. For such reasons, this methodology included the development of a comprehensive ramp identification algorithm to select significant ramp events from the data record, and particular attention was paid to the error analysis during these events. The algorithm allows user input to select ramps of any desired magnitude, and also performs correlation analysis between forecasted ramp events and actual ramp events that coincide within a desired timing window. From this procedure, an investigation of the magnitude and phase of forecast errors was conducted for various forecast horizons. The metrics found to be of most importance for error characterization were selected based on overall impacts, and were ranked in a rudimentary (and perhaps subjective) order of significance. These metrics included: mean absolute error, root mean square error, average magnitude of step changes, standard deviation of step changes, mean bias levels, correlation coefficient of power values, mean temporal bias of ramp events, and others. While these metrics were selected and the methodology was developed for a single dataset, the entire process can be applied generally to any wind power and forecast time series. The implications for such a process include use for generating a synthetic wind power forecast for wind integration studies that will reproduce the same error trends as those found in a real forecast.

In wind integration studies, accurate representations of the wind power output from potential wind power plants and corresponding representations of wind power forecasts are needed, and typically used in a production cost simulation. Two methods for generating “synthetic” wind power forecasts that capture the statistical trends and characteristics found in commercial forecasting techniques are presented. These two methods are based on auto-regressive moving average (ARMA) models and the Markov random walk method. Statistical criteria are suggested for evaluation of wind power forecast performance, and both synthetic forecast methods proposed are evaluated quantitatively and qualitatively. The forecast performance is then compared with a commercial forecast used for an operational wind power plant in the Northwestern United States evaluated using the same statistical performance measures. These quantitative evaluation parameters are monitored during specific months of the year, during rapid ramping events, and at all times. The best ARMA based models failed to replicate the auto-regressive decay of forecast errors associated with commercial forecasts. A modification to the Markov method, consisting of adding a dimension to the state transition array, allowed the forecast time series to depend on multiple inputs. This improvement lowered the artificial variability in the original time series. The overall performance of this method was better than for the ARMA based models, and provides a suitable technique for use in creating a synthetic wind forecast for a wind integration study.

Over the last three decades, significant research has been carried out in the field of wind power forecasting with operational tools, however, making their presence noticeable in the last 15 years. So far, most of the work done has focused on short term forecasting of wind conditions. This is mainly due to the operational need for trading electricity a few days or a few hours ahead of gate closure due to the daily fluctuating nature of the demand and the finite response time of generation plant. However, System Operators (SOs), generators and suppliers have a need for longer term predictions of the power traded in order to maximize financial profits, schedule maintenance, etc. This paper presents a time series analysis of historical observations of wind speed in order to project future wind speed trends. The results suggest that the seasonal trend in wind speeds is the most important factor but that there is some monthly autocorrelation in the data which can improve forecasts. The approach proposed for forecasting wind speeds a month ahead may be deemed useful to suppliers for purchasing base load in advance and to SOs for power systems maintenance scheduling up to a month ahead.

Structural health monitoring is a technique devised to monitor the structural conditions of a system in an attempt to take corrective measures before the system fails. A passive structural health monitoring technique is presented, which serves to leverage historic time series data in order to both detect and localize damage on a wind turbine blade aerodynamic model. First, vibration signals from the healthy system are recorded for various input conditions. The data is normalized and auto-regressive (AR) coefficients are determined in order to uniquely identify the normal behavior of the system for each input condition. This data is then stored in a healthy state database. When the structural condition of the system is unknown the vibration signals are acquired, normalized and identified by their AR coefficients. Damage is detected through the residual error which is calculated as the difference between the AR coefficients of the unknown and healthy structural conditions. This technique is tailored for wind turbines and the application of this approach is demonstrated in a wind tunnel using a small turbine blade held with four springs to create a dual degree-of-freedom system. The vibration signals from this system are characterized by free-stream speed. Damage is replicated through mass addition on each of the blades ends and is located by an increase in residual error from the accelerometer mounted closest to the damaged area. The outlined procedure and demonstration illustrate a single stage structural health monitoring technique that, when applied on a large scale, can avoid catastrophic turbine disasters and work to effectively reduce the maintenance costs and downtime of wind farm operations.

In the early stages of hybrid power system design, it is beneficial to have means of quickly estimating the performance of candidate systems in order to make trades between the components of the system, to incorporate changing requirements, or to deal with uncertainty in assumptions. If the candidate systems incorporate energy storage, the behavior of the system is complicated by the nature of load/renewable/storage interactions, which motivates the use of time-series simulations to accurately evaluate the performance of a candidate system. However, the execution time of a simulation does not lend itself well to real-time engineering design studies. An approach is demonstrated which uses Neural Network surrogate models created to represent a more complex simulation code to enable rapid trades and sensitivities, instant performance results and visualizations, and rapid reoptimization for the design of an on-grid hybrid wind/fossil power system with storage. These surrogate models enabled the creation of a tool that presents a decision maker with the ability to generate endless hybrid mix scenarios and determine which various renewable and non-renewable energy systems meet annual energy load requirements, acquisition and operation costs, and individual solution attributes.