Conference Agenda

Improving the efficiency of heat pumps while using refrigerants with low global warming potential is at the heart of the transformation of the HVAC industry. To address these challenges, system simulation provides reliable data at every step of the design phase, from system architecture evaluation to heat exchanger sizing or virtual control system validation.

Thermodynamic heating and cooling target cycles are derived from power and temperature requirements. During this static analysis, different refrigerants can be explored. These cycles then provide the nominal conditions that determine the sizing of each component of the system using steady-state simulations. Particular attention is paid to heat exchanger models where appropriate heat exchange correlations are essential to ensure the predictivity of the results and where sufficient discretization allows a thorough understanding of the phase change. Through careful examination of local void fractions and slip ratios, valuable information on the refrigerant charge distribution within the system is collected and studied. System performance is ensured by one or more electronic expansion valves whose control strategy is validated using dynamic simulations.

Finally, this complete closed-loop heat pump model is used as a virtual testbed to run hundreds of simulations under all possible external conditions to extract performance maps even before prototyping.

Virtual product design (VPD) of the refrigeration cycle has long been a major subject of many of engineers and manufacturers. Issues should be the question of how simulation gets close to the real world by virtual ways. It is especially true when we consider variable refrigerant flow (VRF) heat pumps, having various types of indoor unit combinations, long and complicated piping configuration, simultaneous cooling and heating functions under extreme weather conditions.

The present study deals with the virtual model development for heat recovery VRF heat pumps aiming at reproducing the dynamic behavior of the prouct in both qualitative and quantitative ways and, thus, saving time and cost caused by the experimental validation during the model development and validation.

Toward this end, we used Modelica, an open platform, object oriented modeling language as the backbone of the virtual model. Dymola, a commercial Modelica compiler, is adopted in order to use specialized HVAC libraries. The incorporation of control algorithm and inverter models is
done in the form of FMU (Functional Mock-up Units). Testing and validaiton of control HW/SW is done in ANSYS Twin-builder, where heat pump dymola model is changed to be FMU in this case.

The choice of open platform SW enhanced synergetic effect among various teams responsible for the components of VRF heat pump system, including cycle model, control algorithm and inverter, and motor models. In the study, the authors share their experience in the virtual design for a VRF heat pump system including the architecture, multiple physics, and the connection with the control SW.

Developed virtual model is validated against measured data, showing it can reproduces all the essential physics of interest both in qualitative and quantitative ways.

The primary aim of this work is to introduce an efficient model for simulating refrigeration units designed for servers and computers. In this model, the unit is conceptualized as a collection of interconnected boxes with the ability to thermally interact with each other.

Each individual box is addressed through the application of a 0D transient energy equation. This equation incorporates factors such as direct solar radiation, thermal emissivity, and convection both to the exterior and to other internal boxes. The model also takes fluid mass transfer into account. To calculate convection coefficients, empirical correlations are used, and external temperature and solar radiation values are sourced from meteorological data specific to the location where the unit is being modeled. The walls of the boxes can be represented either by a thermal resistance or through the resolution of a one-dimensional discretization. Servers are characterized as heat sources with the flexibility to exhibit time-dependent behavior.

To simulate the behavior of the refrigeration unit accurately, an ON/OFF control system with an objective temperature is programmed into the model. Alternatively, other control mechanisms, such as a frequency variator, can be employed for different scenarios.

The model yields comprehensive insights, including temperature profiles over time, heat balances across the walls, and refrigeration requirements as functions of time and seasonal periods.

The proposed framework provides a valuable tool for gaining a deeper understanding of the heat gains and cooling processes associated with servers or other units sharing similar geometries that are exposed to exterior conditions. Subsequently, the model can be employed to investigate the impact and extent of optimizing the unit through various approaches. These may include isolating specific regions, adjusting control parameters, and applying external wall coatings with reflective materials, among other strategies.

Data-driven dynamic models typically offer faster execution than their physics-based counterparts described by large systems of nonlinear and stiff differential-algebraic equations with satisfactory accuracy. Therefore, development of accurate but computationally efficient models directly identified from data constitutes a solution path for investigating controls, fault detection and diagnostics of vapor compression systems (VCS). A modular approach of generating and interconnecting data-driven component models enables reuse of readily trained models and adaption to arbitrary system configurations. Despite the flexibility, a modular integration for system model generation can suffer from nonphysical behaviors of violating conservation laws due to inevitable prediction errors associated with each component model. This paper presents a data-driven dynamic modeling approach that exploits state-of-the-art deep learning methods for constructing component models while enforcing physical conservation for system simulations. Specifically, gated recurrent unit (GRU) and feedforward neural network models are employed for heat exchangers and mass-flow devices. Predictive capabilities and conservation properties of the proposed modeling approach is demonstrated via a case study of a heat pump system. Simulation results reveal a significant speedup with negligible discrepancies compared to high-fidelity physics-based models.

Compressed air pipelines are broadly used in factories to supply pneumatic power to a variety of machineries, sprays, tools, high-precision tables, and many other equipment. Air compressors, however, are power intensive with significant electricity cost. As manufacturing electricity rate depends both on total consumption as well as peak usage, there is an interest to develop forecasting models that may help identify peak usage rather than purely instantaneous demand matching. Also, air compressor power usage trends could be used for anomaly detection and predictive maintenance of the pumps. The compressed air distribution system is often very complex and impossible to develop complete, fluid-dynamic models for it. Recent technologies allow to place IoT pressure and flow sensors on a Wi-Fi network to measure the real time status of a few pipeline branches and terminals. Still, managing the compressed air system is not simple due to the changes in operation schedule in different areas of the factory. Here we propose a dynamic, compact circuit model for the compressed air distribution system that considers different tube diameters, lengths, storage tanks and usage flows. The goal is not to precisely model the geometry of compressed air distribution, but to see if the historical, transient power trace could be used to identify main usages that can then be further correlated with the production schedule. We first analyze air compressors’ power time series over a couple of years. We extracted the daily, weekly, and seasonal characteristic variations using a spectral transform technique. Based on the observations, we developed a compact analytic model for parametric simulations using an analogy with an electrical network. Nonlinear and transient behavior of the air movement across the pipeline was modeled based on the acoustic system. When time goes to infinity, the pressure distribution follows a static, fluid-dynamic system in steady-state. Then, we calculate the time dependent response using an electrical circuit simulator “SPICE”. A simple toy model demonstrates that the compressor electrical power consumption signals contain the characteristic information of pneumatic terminals with given time constants. The response time can be estimated by the analytic reduced-order model. We found that power consumption time series with sub-seconds resolution is required to identify the distribution pattern in a typical factory with thousands of feet of compressed air piping. Future work will focus on correlating transient power-consumption patterns with the production or usage schedule.

First-principles dynamic models of vapor compression cycles (VCCs) have demonstrated promising capabilities in capturing complicated thermo-fluid component and system behavior. Development and calibration of these models for achieving reliable predictions is of critical importance for applications to control design and fault detection and diagnostics. Nevertheless, the inherent complexity of model descriptions by large systems of differential-algebraic equations presents significant challenges in model calibration processes that utilize classical gradient-based methods to minimize discrepancies between model predictions and available measurements. This paper presents a reduced-dimension Bayesian optimization (BO) framework for calibrating transient VCC models that employs a low-dimensional probabilistic surrogate model to approximate an expensive-to-evaluate calibration cost associated with each set of candidate parameters, and optimal search over a feasible parameter space. The proposed approach was implemented for calibrating a large set of parameters including heat transfer coefficients and component geometries for an air-source heat pump based on laboratory measurements. The optimal set of calibrated parameters yielded exceptional prediction accuracy compared against measurements and original models without calibration.