The performance of HVAC equipment, including chillers, is continuing to be pushed to theoretical limits, which impacts the necessity for advanced control logic to operate them efficiently and robustly. At the same time, their architectures are becoming more complex; many systems have multiple compressors, expansion devices, evaporators, circuits, or other elements that challenge control design and resulting performance. In order to maintain respectful controlled speed of response, stability, and robustness, controllers are becoming more complex, including the move from thermostatic control, to proportional integrator (PI), and to multiple-input multiple-output (MIMO) controllers. Model-based control design works well for their synthesis, while having accurate models for numerous product variants is unrealistic, often leading to very conservative designs. To address this, we propose and demonstrate a learning-based control tuner that supports the design of MIMO decoupling PI controllers using online information to adapt controller coefficients from an initial guess during commissioning or operation. The approach is tested on a physics-based model of a water-cooled screw chiller. The method is able to find a controller that performs better than a nominal controller (two single PI controllers) in terms of decreasing deviations from the operating point during disturbances while still following reference changes.

Calibrated energy models are used for measurement and verification of building retrofit projects, predictions of savings from energy conservation measures, and commissioning building systems (both prior to occupancy and during real-time model based performance monitoring, controls and diagnostics). This paper presents a systematic and automated way to calibrate a building energy model. Efficient parameter sampling is used to analyze more than two thousand model parameters and identify which of these are critical (most important) for model tuning. The parameters that most affect the building’s energy end-use are selected and automatically refined to calibrate the model by applying an analytic meta-model based optimization. Real-time data from an office building, including weather and energy meter data in 2010, was used for the model calibration, while 2011 data was used for the model verification. The modeling process, calibration and verification results, as well as implementation issues encountered throughout the model calibration process from a user’s perspective are discussed. The total facility and plug electricity consumption predictions from the calibrated model match the actual measured monthly data within ±5%. The calibrated model gives 2.80% of Coefficient of Variation of Root Mean Squared Error (CV (RMSE)) and −2.31% of Normalized Mean Bias Error (NMBE) for the whole building monthly electricity use, which is acceptable based on the ASHRAE Guideline 14–2002. In this work we use EnergyPlus as a modeling tool, while the method can be used with other modeling tools equally as well.

Building recommissioning in essential in the aging building stock to maintain efficient and comfortable operation as equipment ages and portions of the building are re-purposed for uses other than what was originally intended. Model-based recommissioning provides a way to evaluate payback and incentives for equipment replacement, and the response of the building to optimized operational strategies - without disturbing the comfort or productivity of current occupants. Accurate models are needed for these investigations, which must be calibrated to available sensor data. In this work, cheap wireless temperature transmitters are installed in a 40 year old building to gain information about the performance of the buildings envelope for model calibration purposes. The goal is to have a calibrated energy model that is sufficient for control system optimization. To perform this task, a detailed model is built and envelope parameters of this model are calibrated by using feedback control to identify components of zone heat balances. Uncertainty and sensitivity analysis along with optimization is used to calibrate envelope parameters so that heat transfer through the envelope is captured accurately.

Most people are intimately involved with the built environment, while unfamiliar with detailed aspects of its design and operation. Buildings are everywhere and are designed and equipped using an agglomeration of many different design elements or puzzle pieces. With the recent trends towards a more energy efficient world, there has been an attempt to make buildings more efficient by using highly efficient pieces of the design puzzle. Not always does the integration of these subsystems result in an efficient building as a whole. The goal of this chapter is to highlight some of these boundaries and current engineering trends to surpass these obstacles. The discussion will be focused on large commercial buildings in the United States, while similar concerns are prevalent in other building types and in other global locations. We start by highlighting different ways that performance is measured and review the different design elements and equipment choices that are available to construct a building. The large number of interacting components creates complexity and a challenge to obtain a high performance structure. Specifically, technology barriers to realizing high performance buildings through this integration process lie in the ability to create useful models, data analysis tools, and effective control strategies. The chapter concludes with some current applied research in building systems that address the complexities in building systems and methods being developed to overcome the barriers that lie in the way.

As building energy modelling becomes more sophisticated, the amount of user input and the number of parameters used to define the models continue to grow. There are numerous sources of uncertainty in these parameters, especially when the modelling process is being performed before construction and commissioning. Past efforts to perform sensitivity and uncertainty analysis have focused on tens of parameters, while in this work, we increase the size of analysis by two orders of magnitude (by studying the influence of about 1000 parameters). We extend traditional sensitivity analysis in order to decompose the pathway as uncertainty flows through the dynamics, which identifies which internal or intermediate processes transmit the most uncertainty to the final output. We present these results as a method that is applicable to many different modelling tools, and demonstrate its applicability on an example EnergyPlus model.

Whole-building energy models are used in practice to predict energy and comfort for an entire building given its architectural and built state, external stimuli from weather; and internal behavior of both the equipment and occupants of the building. There exists both open source and commercial software for simulating such cases for an entire year at sub-hourly reporting intervals. Unfortunately, the dynamics of the building are masked in assumptions included in the numerical routines that are often intertwined within the thermal physics. Because of this, control-oriented analysis is limited to performing exhaustive time-based simulations. In this paper, we describe a method to analytically extract the dynamics from a whole-building energy simulator for the purpose of control and dynamical systems analysis. In this way, the function of the energy simulator is only a user interface and a means to organize information inherent in the dynamics (capacitances, interaction between elements of the building, etc.). We provide a test case on a medium office building and illustrate some of its control-oriented dynamic properties using EnergyPlus as the simulator.

: Retrofitting the existing building stock represents the largest and fastest way to reduce energy consumption for the DoD. However the current retrofit delivery process is manually intensive and expensive, focused on equipment selection for initial cost and not energy performance, and the design tools are not amenable to systems solutions that have the potential for substantially reducing energy consumption in buildings. Systems methodology and tools are necessary to deliver deep retrofits, i.e. significantly higher energy performance in existing buildings than is achievable by the current retrofit process. The report describes newly developed screening methodology and tools for early assessment of deep retrofit potential across the entire DoD stock of 250,000 buildings, use of sensitivity and uncertainty analysis tools to isolate critical design parameters and establish performance bounds during design, and reduced-order modeling tools for highly energy efficient building system control design. Validated tools were developed, and retrofit system options that can reduce energy consumption by 30-50% have been identified with existing DoD building use cases.

This article presents the calibration of a building energy model of a historic office building, developed using the EnergyPlus simulation program. The building under study is the Fleet and Family Support Center located at Naval Station Great Lakes, IL. It was built in 1901 and renovated multiple times. This building has a total floor area of 36,843 ft(2) (3424 m(2)) and mainly consists of offices and conference rooms. An extensive sensitivity study that efficiently perturbs more than two thousand model parameters is employed for model calibration. Those parameters that most affect the building's energy end-use are selected and automatically refined to calibrate the model by applying an analytic meta-model based optimization. Real time data including weather and energy meter data in 2010 is used for model calibration and 2011 data is used for model verification. The modeling process, calibration and verification results, as well as implementation issues encountered throughout the model calibration process from a user's perspective are discussed. The total facility and plug electricity consumption predictions from the calibrated EnergyPlus model match the actual measured monthly data within +/-5%.

Building failure modes can be defined as the range of possible faults, mistuning, degradation and wear that can occur in building systems. They drive poor operational energy performance that can be achieved as compared to what was expected in design. If the more critical failure modes can be uncovered early in design, they can be mitigated through design changes or monitoring of associated characteristics on parameters such as temperatures, flow rates and pressures. To understand and prioritize failure modes, failures are characterized through mappings to whole building simulation input variables, and then random sampling methods are applied to simulate the building operating under combinations of different failure modes. Reduced order models are then fit to the resulting data to detemine which failure modes are more critical. The approach is demonstrated on a mid-sized office building, comparing a standard and advanced retrofit design.

Whole-building energy models take information about the structure of a building, its equipment (electrical loads, lights, conditioning equipment, etc.), and disturbances (people, weather) and predict its year long comfort and energy performance. Both commercial and freely available tools are available for performing these time-domain simulations, which are used for design trade studies and more frequently to check for energy consumption and comfort compliance. These models require hundreds of assumptions as input when it comes to parameterizing the building model. Previous studies have investigated how predictions are influenced by these assumptions and which of the parameters are critical to year-long calculations. In this paper we extend this approach to investigate how parametric uncertainty influences uncertainty in the energy dynamics within a building. We provide a case study that investigates an office building by extracting dynamic information out of an EnergyPlus model, and supplies this information to an automatically generated analytical thermal network model. We conclude with a control-oriented frequency-based robustness assessment as well as a study of how uncertainty influences the network structure of the building by investigating the spectral gap of its graph Laplacian.

Calibrated energy models are useful for commissioning building systems, measurement and verification (M&V) of building retrofit projects, and predictions of savings from energy conservation measures. This paper presents the modeling and calibration process for building energy models of a DoD (Department of Defense) building. The models are developed using EnergyPlus and TRNSYS simulation programs with measured data from an enhanced building management system (BMS) which includes an on-site weather station. The building under study is the Atlantic Fleet Drill Hall located at Naval Station Great Lakes, IL. This LEED (R) Gold certified building with a total floor area of 69,218 ft(2) (6,431 m(2)) consists of a drill deck and administrative offices. Static data from as-built drawings and dynamic data from building operations are collected and analyzed to create energy models with EnergyPlus and TRNSYS. An extensive sensitivity study by systematically perturbing more than a thousand model input parameters is employed for model calibration. Those parameters that most affect the building energy end-use are selected and refined to calibrate the models. The calibration results, as well as problems encountered throughout the process from the user's perspective, are discussed. The total facility and individual equipment electricity consumption predictions from the calibrated models closely match the measured data.

As building energy models become more accurate and numerically efficient, model-based optimization of building design and operation is becoming more practical. The state-of-the-art typically couples an optimizer with a building energy model which tends to be time consuming and often leads to suboptimal results because of the mathematical properties of the energy model. To mitigate this issue, we present an approach that begins by sampling the parameter space of the building model around its baseline. An analytical meta-model is then fit to this data and optimization can be performed using different optimization cost functions or optimization algorithms with very little computational effort. Uncertainty and sensitivity analysis is also performed to identify the most influential parameters for the optimization. A case study is explored using an EnergyPlus model of an existing building which contains over 1000 parameters. When using a cost function that penalizes thermal comfort and energy, 45% annual energy reduction is achieved while simultaneously increasing thermal comfort by a factor of two. We compare the optimization using the meta-model approach with an approach using the EnergyPlus model integrated with the optimizer on a smaller problem using only seven optimization parameters illustrating good performance. (C) 2011 Elsevier B.V. All rights reserved.

Large amounts of sensor information is often captured from either realworld building sensors, or virtual building models, for many purposes including control design, fault or aging analysis, and model calibration. Because of the large dimension of this data on both spatial and temporal scales, it is often challenging to come to quick conclu- sions about what information of engineering importance is in the data. In this paper we present an approach to quickly assess spatial information in data based on the spectral content of a certain projection operator. We use operator theoretic methods to capture Koopman modes that represent the spatial content of oscillations in ther- mal quantities. By investigating these modes for different physically significant time-scales (e.g. diurnal, or control system time-scales) we can quickly capture how different parts of a building are responding to load changes at these frequencies (”breathing”). This information helps us to understand anomalies in different aspects of the data, as well as out of phase behavior between zones which may highlight areas of poor control system performance. We present actual and EnergyPlus data from a real building (170K square foot building with approximately 2000 data points) and illustrate how this approach to data analysis and model validation highlights aspects of the data which may otherwise have been overlooked.

Metastable escape is ubiquitous in many physical systems and is becoming a concern in engineering design as these designs (e.g., swarms of vehicles, coupled building energetics, nanoengineering, etc.) become more inspired by dynamics of biological, molecular and other natural systems. In light of this, we study a chain of coupled bistable oscillators which has two global conformations and we investigate how specialized or targeted disturbance is funneled in an inverse energy cascade and ultimately influences the transition process between the conformations. We derive a multiphase averaged approximation to these dynamics which illustrates the influence of actions in modal coordinates on the coarse behavior of this process. An activation condition that predicts how the disturbance influences the rate of transition is then derived. The prediction tools are derived for deterministic dynamics and we also present analogous behavior in the stochastic setting and show a divergence from Kramers activation behavior under targeted activation conditions.