Search

Integrating renewable energy and energy storage systems provides a way of operating the electrical grid system more energy efficiently and stably. Thermal storage and batteries are the most common devices for integration. However, it is not clear which integrated storage system performs better in terms of overall economics. Ice storage has low initial and maintenance costs, but there is an efficiency penalty for charging of storage and it can only shift electrical loads associated with building cooling requirements. A battery's round-trip efficiency, on the contrary, is quite consistent and batteries can be used to shift both HVAC and non-HVAC loads. However, batteries have greater initial costs and a shorter life. This research presents a tool, using model predictive control and optimal sizing, and provides a case study for comparing life-cycle economics of battery and ice storage systems for commercial buildings that have chillers for cooling and an on-site photovoltaic system.

Small and medium-sized commercial buildings (SMCB) are significant demand response resources, and it is important to develop grid-responsive control algorithms that exploit those resources and create financial benefits for building owners and HVAC service providers. Furthermore, unlike large-sized commercial buildings, there is an opportunity to have universally applicable control solutions for many SMCBs since those buildings have a consistent HVAC system configuration: SMCBs are commonly served by multiple-staged air conditioning units controlled by their own thermostats. Despite the demand response potential and scalability, however, very few control solutions are available for SMCBs. Typical model predictive control (MPC) and heuristic control approaches for cooling load shifting that lower thermostat setpoints before an electric price jump are suitable mainly for large-sized commercial buildings where a continuous capacity modulation is possible, e.g., via dampers in variable air volume terminal units. However, those approaches can cause undesired, high peaks for SMCBs due to the nature of ON/OFF unit staging and narrow thermostat deadbands. This could discourage the use of advanced grid-responsive controls for SMCBs due to the concern of high demand charges, and has to be resolved. This paper presents a MPC solution that overcomes this challenge. It has a hierarchical MPC structure where an upper level MPC is responsible for electrical load shifting in response to an electric price signal while a lower level MPC is responsible for coordinating compressor stages to eliminate unnecessary peaks and follows the setpoints determined by the upper level MPC. Two one-month, comprehensive laboratory tests have been carried out to demonstrate load shifting and cost savings for the algorithm. Interesting trade-offs between energy efficiency and load flexibility were observed and are discussed, and lessons learned for applying MPCs for SMCBs are also presented.

A computationally efficient but accurate dynamic modeling approach for vapor compression systems is important for many applications. Nonlinear model order reduction techniques which generate reduced order models based on high fidelity vapor compression cycle (VCC) models are attractive for the purposes. In this paper, a number of technical challenges of applying model order reduction methods to VCCs are described and corresponding solution approaches are presented. It starts with a reformulation of a standard finite volume heat exchanger model for matching the baseline model reduction structure. Reduced order models for evaporator and condenser are constructed from numerical snapshots of the high fidelity models using Proper Orthogonal Decomposition (POD). Methodologies for system stability and numerical efficiency of POD reduced order models are described. The reduced order heat exchanger models are then coupled with quasi-static models of other components to form a reduced order cycle model. Transient simulations were conducted over a wide range of operating conditions and results were compared with the full order model as well as measurements. The validation results indicate that the reduced order model can execute much faster than a high-fidelity finite volume model with negligible prediction errors.

Previous mode switching algorithms for heat exchanger moving boundary models in the literature are composed of a set of IF-THEN rules. These representations could lead to numerical challenges due to the inherited discontinuities associated with IF-THEN rules. This paper presents an alternative mode switching methodology which results in a continuous moving boundary heat exchanger model over all possible mode changes. Numerical performance of the proposed method for a heat exchanger has been tested using simulations and sample results are compared with a moving boundary model with switching based on IF-THEN rules and also a finite-volume method. The proposed switching algorithm was implemented within a complete vapor compression cycle model and results were compared with experimental data for a start-up transient.

Developing a computationally efficient but accurate building energy simulation (BES) model is important for many purposes. Model order reduction (MOR) methods are attractive and much more reliable than identification approaches, since it directly extract a lower-dimensional model from a detailed physics-based model without any pre-simulations. However, because of computational and data storage requirements, there are challenges of applying these methods to a large-scale building. To overcome the problem, this paper introduces the Krylov subspace method to the building science field. Technical issues of applying the method to building applications are addressed and a suitable algorithm that overcomes those challenges is presented. To demonstrate the reliability of the algorithm, comparisons between the resulted reduced-order model (ROM) and a high-fidelity model from a commercial BES software for a 60-zone case study building are provided. The ROM was a factor of 100 faster than the high fidelity model but with high accuracy.

Growth of a frost layer on an evaporator surface due to low evaporator temperature as well as moisture contained in surrounding air deteriorates performance of a refrigeration system significantly and requires significant energy for defrost. Many studies have been performed to model the heat and mass transfer phenomena in an attempt to have insight and accurate prediction. However, many models form nonlinear algebraic differential equations which require iterative numerical solvers. Computationally efficient but accurate models are needed in order to evaluate overall system performance. The objective of this paper is to introduce a modeling approach to overcome the problem. A solution strategy based on an enthalpy-based reformulation and linearization method will be presented. Comparisons of the proposed and detailed model results for both flat plate and finned tube heat exchangers are provided. The proposed modeling approach is around 10 times faster than reference models while maintaining comparable accuracy.

Small and medium sized commercial buildings, such as retail stores, restaurants and factories, often utilize multiple packaged air conditioners, i.e. roof top units (RTUs), to provide cooling and heating for open spaces. A conventional control approach for these buildings relies on local feedback control, where each unit is cycled on and off using its own thermostat. The lack of coordination between RTUs represents a missed opportunity for operating more efficient units when there is strong coupling between the spaces they serve and can lead to unnecessarily high electrical demand due to the inherent randomness of unit cycling. This paper presents an overall model-based predictive control (MPC) approach for RTU coordination that includes a description of the control architecture, modeling approach, implementation, and assessment. We provide results of laboratory and field tests that demonstrate the short-term and long-term performance of the MPC solution in terms of energy and demand savings.

It is important to have practical methods for constructing and learning a good mathematical model for a building's thermal system in the presence of unmeasured disturbances and using data from closed loop operation. With this goal in mind, this paper presents a mathematical framework that explains the asymptotic behavior of an estimated model under those conditions and that can aid in learning an accurate model. Some analytic results from the literature of system identification are extended and interpreted for building systems. A new identification approach for determining an accurate thermal network (RC) model for a multi-zone building is developed based on the analytic result, and its superior performance over a conventional grey-box modeling approach is demonstrated experimentally.

This report presents results from field testing and comfort surveys designed to evaluate peak demand-limiting strategies that utilize both precooling and adjustments of zone cooling setpoints. The testing was performed over a two-week period at a small bank building in Palm Desert, California. During the first week test, three kinds of control strategies were considered: 1) conventional night setup control as a baseline case, 2) a simple linear-rise demand-limiting strategy that involved precooling during the morning and linear setpoint adjustments during an afternoon demand-limiting period, and 3) a simple step-up demand-limiting strategy that included precooling in the morning and resetting of setpoint during the demand-limiting period.During the second week of testing, a demand-limiting strategy was tested for four days with setpoint trajectories determined using a weighted-averaging method developed at Purdue University. Precooling of the building was performed at 70ºF setpoint from 6am to 12pm and setpoints during the on-peak period from 12pm to 6pm were modulated from 70 to 78ºF following a trajectory that attempted to minimize peak cooling load. (The measured temperature at the polling station was a minimum of 1.5 degrees F above the thermostat setpoint (see figures 24 to 29 in appendix). The baseline was conventional night-setup control with a 72ºF cooling setpoint temperature during the occupied period. The demand-limiting tests resulted in greater than 30% reduction of peak air conditioner power on average for the four tested days which accounted for 0.76W/ft2 peak savings. The comfort survey revealed that the response of occupants was highly variable at any given indoor temperature. Statistical analysis of all the data collected, including baseline days and test days, indicated a significant probability that a given occupant will vote that the temperature is ‘cool’ at the low setpoint temperature of 70 degrees (between 30 and 50 percent), and ‘warm’ at the upper setpoint of 78 degrees (between 37 and 52 percent) (figure 21). However, only half of these votes are at the level where the ii respondent says it ‘bothers’ them. The probability of a given occupant being bothered by the ‘cool’ temperature at the low setpoint is estimated to be 17 percent and the probability of a given occupant being bothered by the ‘warm’ temperature at at the upper setpoint is estimated to be 23 percent. (figure 22). If we assume the neutral temperature to be between 74 and 75 degrees F, the probability of dissatisfaction (both ‘Too warm! It bothers me’ and ‘Too cool! It bothers me’) is estimated to be 20 percent.

Most commercial buildings have been dedicated HVAC systems to meet their comfort needs. In larger commercial buildings, the control of this equipment is achieved through use of computerized control systems, which allow the flexibility of running the building under different control strategies. Most control systems vary the building temperatures using a strategy called night-setup control. Under night-setup control, the air temperatures are in the middle of the comfort range and the cooling equipment is usually active during daytime hours. During the night and weekends, a high temperature setpoint is sent out by the control system and the cooling system is typically inactive during these periods. However, recently there has been an increased emphasis on developing dynamic building control strategies, which attempt to minimize the total cooling costs of a building and reduce peak cooling demand. Accurate building modeling tools are needed for predicting the building thermal loads under different control strategies. The work described in this report involved development and evaluation of models using data from the Energy Resource Station (ERS) at the Iowa Energy Center (IEC). Models trained with data from the ERS were used to estimate peak cooling load reduction associated with different demand-limiting and precooling strategies. The IEC is a relatively lightweight structure that is representative of small commercial buildings. Demand reduction results for this building would be smaller than those possible for larger commercial buildings. The specific work tasks that provided the basis for the results described in this report are summarized as follows: A. Develop forward simulation model for the ERS building B. Develop inverse simulation model for the ERS building C. Evaluate demand-limiting control strategies