Proceedings of IEEE () 2015 - p. , ISSN
http://arxiv.org/abs/1505.00078

Co-simulation platforms are necessary to study the interactions of complex systems integrated in future smart grids. The Virtual Grid Integration Laboratory (VirGIL) is a modular co-simulation platform designed to study interactions between demand response strategies, building comfort, communication networks, and power system operation. This paper presents the coupling of power systems, buildings, communications and control under a master algorithm. There are two objectives. First, to use a modular architecture for VirGIL, based on the Functional Mock-up Interface (FMI), where several different modules can be added, exchanged, and tested. Second, to use a commercial power system simulation platform, familiar to power system operators, such as DIgSILENT Powerfactory. This will help reduce the barriers to the industry for adopting such platforms, investigate and subsequently deploy demand response strategies in their daily operation. VirGIL further introduces the integration of the Quantized State System (QSS) methods for simulation in this co-simulation platform. Results on how these systems interact using a real network and consumption data are also presented.

Energy and Buildings () 2015 - p. , ISSN 0378-7788
http://dx.doi.org/10.1016/j.enbuild.2015.10.017.

Most of the state-of-the-art whole building simulation programs formulate models using imperative programming languages. This results in programs with procedural computer assignments that change the state of the program and describe how the program operates. This approach has its origin in the seventies and complicates modeling and use of efficient methods for simulation and optimization. In contrast, formulating models using an equation-based language that declares relations among variables rather than prescribing computing instructions allows models to be analyzed and manipulated automatically using computer algebra to generate efficient code for simulation or optimization. In this paper, we contrast the new approach to the state-of-the-art in building energy simulation. We explain how such manipulations support new use cases for building energy simulation. As applications, we first describe how to translate an EnergyPlus model into an equation-based model. We then couple this model to models of electrical grid, HVAC and controllers for demand response and reactive power control to study the effect on power quality using a coupled simulation of all domains. In this example, the combined building room temperature and photovoltaic inverter control ensures that the voltage stays within the admissible limits. In a second example, we contrast the computing performance for solving an optimal control problem for a room-level model predictive controller with and without symbolic manipulations. In this example, symbolic manipulations led to a 1,000 times faster solution.

Building Simulation (7) 2014 - p. 439-454, ISSN 1996-3599
http://dx.doi.org/10.1007/s12273-014-0175-6

Airflow modelling is of fundamental importance for evaluating ventilation performance and energy consumption in buildings, and various approaches to the problem—starting from purely empirical up to the CFD ones—have been proposed and evaluated in the past years. Moreover, since the ultimate goal is whole building modelling, airflow simulation needs coupling with Energy Simulation (ES), in order to assess the overall energy performance. Due to the substantial differences between the software employed for airflow and ES, co-simulation is very often felt as the only way to handle such a problem. For example, in recent years a lot of effort has been spent in to couple ES and CFD tools. This paper proposes an alternative, in the form of an approach for solving the Navier-Stokes equations in a general multi-domain modelling framework. Since co-simulation is not involved, the correctness of the numerical solution relies on a single solver, thus being really transparent to the analyst. This is a first step towards a whole building simulation tool embedded in a unique framework capable of performing energy analysis, computing airflows, and representing control systems.

Applied Energy (124) 2014 - p. 156 - 166, ISSN 0306-2619
http://dx.doi.org/10.1016/j.apenergy.2014.03.009

This work presents a robust and computationally efficient algorithm for both whole-building and component-level energy fault detection and diagnosis (FDD). The algorithm is able to provide reliable estimation of multiple and simultaneous fault conditions, even in the presence of noisy and sometimes erroneous sensor data, and to provide uncertainty estimation. The algorithm can be used to provide such outputs as the probability of a fault, the likely cause(s), and the expected consequences of the fault(s) on energy use. The approach is based on an advanced Bayesian nonlinear state estimation technique called Unscented Kalman Filtering, but with our addition of a back-smoothing method that provides fast and robust FDD for common building use cases. The approach is presented and demonstrated for detecting energy and hydraulic faults in a chiller plant. The model of the chiller plant is a subsystem of an actual chiller plant, calibrated to real data. The algorithm can detect common faults, such as (1) energy faults (e.g., the chiller is not working properly, or far from its nominal condition), (2) functional faults caused by issues in the compressor and (3) occlusions in the valves that may reduce the water flow rate through the condenser and evaporator water loop. It is also shown that estimates of uncertainty are consistent with the error in the synthetic data, and can be updated as new data stream in from sensors.

Journal of Process Control (22) 2012 - p. 1237 - 1245, ISSN 0959-1524
http://dx.doi.org/10.1016/j.jprocont.2012.05.012

A previous paper addressed the “PI+p” controller (a PI augmented with an additional, stable pole) and demonstrated its advantages over standard PIs, particularly if tuned with a combined model-relay-based method along the “contextual” approach proposed in another recent work. This manuscript extends the scenario to the PID+p structure, with obvious meaning, and applies the contextual approach to its autotuning. Benchmark tests and a quite realistic simulation example support the proposal.

Simulation Modelling Practice and Theory (22) 2012 - p. 1 - 12, ISSN 1569-190X
http://dx.doi.org/10.1016/j.simpat.2011.11.001

This manuscript presents a quasi-3D modelling approach for the thermo-hydraulic simulation of airflows. The approach is tailored to low velocity motion fields, thus suited for building rooms, household refrigerator cavities, and so forth. Its goal is to somehow bridge the gap between fully mixed volumes, often too inaccurate for system-level studies, and Computational Fluid Dynamics (CFDs) models, that for the same purpose are frequently too demanding from the computational standpoint. The proposed approach allows to conduct informative and scalable-detail studies, including relevant facts such as air movers’, heaters’ and sensors’ positioning, without resorting to co-simulation, to the advantage of engineering efficiency.

Mathematical and Computer Modelling of Dynamical Systems (17) 2011 - p. 543-559, ISSN
http://dx.doi.org/10.1080/13873954.2011.592143

When modelling and simulating a building to evaluate its energy-related performance, reproducing the behaviour of the contained air volumes is very important. For that purpose, fully mixed models (i.e. for instance, a single temperature per room) easily prove inadequate, while computational fluid dynamics ones do not provide sufficient simulation speed, and are difficult to formulate and manage in a modular manner. This manuscript presents a somehow intermediate solution based on the object-oriented modelling paradigm, and implemented in the Modelica language. The proposal advantages are illustrated by two significant examples.

The IEA EBC Annex 60 is developing and demonstrating new generation computational tools for building and community energy systems based on the non-proprietary Modelica modelling language and Functional Mock-up Interface standards. The anticipated outcomes are open-source, freely available, documented, validated and verified computational tools that allow buildings, building systems and community energy grids to be designed and operated as integrated, robust, performance based systems with low energy use and low peak power demand. Activity 2.3 focuses on the use of models during building operations to augment monitoring, implement control algorithms and fault detection, and diagnostics methods. This paper presents an overview of the work carried out within Activity 2.3 including a description of three case studies.

Residential and commercial buildings consume nearly 75% of the overall electrical energy in the U.S., and the amount of renewable energy in the grid keeps increasing. As buildings can shift their loads to reduce peak demand and to flatten the ramps of load increase and decrease, buildings are an important contributor to ensuring a stable grid operation. Simulation models that account for building loads and building dynamics, as well as their impact on the electricity distribution grid, are essential to assess different design and control options for buildings and electrical systems. This paper presents the use of such simulation models of dynamic building loads coupled to electrical models within a tool chain that allows efficient numerical solution of nonlinear optimization problems that aim at controlling the voltage stability and thermal comfort while minimizing energy consumption or energy cost. The optimization problems are formulated using the open-source JModelica software which brings them to a form in which they are solved using the nonlinear programming solver IPOPT. In this formulation, JModelica converts an infinite-dimensional optimization problem, defined on the solution to the equations of the simulation model, to a finite-dimensional nonlinear programming problem using computer algebra and collocation methods.

Besides the building model, state estimation is a crucial part of model predictive control for buildings. The main reasons are the scarcity of measurements, their greatly varying quality, and the inability to measure all the state variables, e.g., internal mass temperature. This paper compares three different state estimation techniques and shows how they can improve the ability of the building model to correctly predict the behaviour of the real building. We compare an unscented Kalman filter, a deterministic approach and a moving horizon estimation scheme. The methods are compared for their ability to estimate unmeasured state variables, and the effects that the estimations have on the model’s ability to predict the dynamics of the building. We tested the methods with both virtual simulation data and real measurement data for a multizone office building in Brussels.

This paper presents an update about recent additions to the Modelica Buildings library. The Modelica Buildings library is a free open-source library with dynamic simulation models for building energy and control systems. The library contains models for air-based HVAC systems, chilled water plants, water-based heating systems, controls, room and envelope heat transfer, multizone airflow, including natural ventilation and contaminant transport, and electrical systems. The recent major additions that will be presented in this paper are a CFD model that is embedded in a thermal zone, a package for buildings to electrical grid integration, models for demand response simulation and data-driven models to predict the building load. The CFD model is an implementation of the Fast Fluid Dynamics code that allows three-dimensional CFD inside a thermal zone, coupled to building heat transfer, HVAC components and feedback control loops. The package for buildings to electrical grid integration includes models for loads, transformers, cables, batteries, PVs and wind turbines. The models can be configured for DC or AC systems with two or three phase balanced and unbalanced systems. They compute voltage, current, and active and reactive power based on the quasi-stationary assumption or using the dynamic phasorial representation. The models for demand response include a demand response client and data-driven models that predict the anticipated load of a building. This paper will explain these packages and present illustrative examples.

Energy simulation programs are increasingly used to assess the energy performance of buildings and heating, ventilation, and air conditioning (HVAC) systems during design. Faults during operation can lead to considerable increase in energy use and associated operating costs. In 2009, 13 of the most common energy faults in U.S. commercial buildings were thought to cause wastes of over $3.3 billion per year. Fault detection and diagnosis (FDD) tools can reduce some of these energy wastes. In the buildings industry there is still a disconnect between the design phase, commissioning, and operation. This work presents a process and a tool chain for model-based real-time automated FDD. It facilitates the introduction of FDD in building applications by allowing simulation models from the design stage to be reused during operation. Our tool chain uses models that comply with the Functional Mockup Interface (FMI) standard, a standard developed by 29 partners since 2008 that is supported by more than 40 simulations tools, including Modelica simulation environments and MATLAB. Engineers that use simulation tools compliant with the FMI standard (Blochwitz, 2011) can reuse their models for creating energy-aware FDD algorithms. This work presents and explains the general software architecture and FDD algorithm, which is based on a hybrid of various existing statistical algorithms: Bayesian updating, Unscented Kalman Filtering, and back-smoothing. We also demonstrate the algorithm to analyze simultaneous faults in a chiller plant.

This work presents a toolchain based on the Functional Mockup Interface (FMI) standard for accelarating the adoption of model-based Fault Detection and Diagnostics (FDD) algorithms. The presented toolchain uses nonlinear state estimation techniques to detect multiple simultaneous faults in presence of noisy sensor measurements. The toolchain enables the use of a dynamic model that emulates correct and faulty equipment, while keeping the computing effort low. The FDD toolchain allows importing dynamic models through an interface that complies with the FMI standard, an application programming interface for the exchange of models which is supported by many simulation tools. This functionality reduces the modeling effort required to setup a model-based FDD algorithm because it allows taking advantage of models available in open-source libraries for building components and HVAC systems that can model various fault scenarios.

This paper presents the new Electrical package of the Modelica Buildings library. The package contains models for different types of sources, loads, storage equipment, and transmission lines for electric power. The package contains models that can be used to represent DC, AC one-phase, and AC-three phases balanced and unbalanced systems. The models have been successfully validated against the IEEE four nodes test feeder. An example shows how the models can be used to study the integration and control of renewable energy sources in a electric distribution grid.

DOI: http://dx.doi.org/10.1109/UCC.2014.85

Cloud computing has recently received considerable attention, as a promising approach for delivering ICT services by improving the utilisation of data centre resources. On the other hand, the increased usage of ICT, jointly with the increased cost of energy, make designing and managing data centres with energy efficient strategies a crucial and strategical problem. In the literature there are many studies based on Computational Fluid Dynamics (CFD) simulation models. While the obtained results are usually very reliable and accurate, the massive use of such models for control design require a large amount of time due to their computational complexity and scarce flexibility. In addition, CFD models are usually quite sensitive to any kind of change to the operating conditions, e.g. When varying the considered load balancing algorithm. In this paper, we propose a modelling and simulation framework suited for control design, that is able to simulate both heating and energy phenomena, and different load balancing strategies in a data centre. The purpose of this work is to provide a tool to perform energy-efficiency targeted studies, providing accurate and reliable results.

DOI: 10.3182/20140824-6-ZA-1003.01409

Desiccant wheels can improve the efficiency of air handling units, as their inherent heat recovery capabilities reduce the external energy needs. On the other hand, however, desiccant wheels introduce state variable interactions that are not present in traditional units. Hence, to actually yield the possible advantages, air handling units with desiccant wheels require accurate control, which in turn calls for dynamic wheel models that can be parametrised from design data, so as to allow for virtual prototyping, and couple reliability with numerical efficiency, to deliver the numerous system-level simulations required with an acceptable computational effort. This paper proposes a dynamic model for desiccant wheels fulfilling the above requirements. To support the proposal, examples ere reported on the use of the proposed model for the design and control of innovative unit layouts.

DOI: 10.3384/ECP14096647

This paper proposes a solution for creating a model-based state and parameter estimator for dynamic systems described using the FMI standard. This work uses a nonlinear state estimation technique called unscented Kalman filter (UKF), together with a smoother that improves the reliability of the estimation. The algorithm can be used to support advanced control techniques (e.g., adaptive control) or for fault detection and diagnostics (FDD). This work extends the capabilities of any modeling framework compliant with the FMI standard version 1.0.

DOI: http://dx.doi.org/10.1109/CCA.2012.6402401

Simulation is widely used to assess and/or improve the energy efficiency of both existing and new buildings. Such an analysis has to account for heterogeneous phenomena efficiently, to manage components in a modular manner, and (which is seldom addressed in a structured way) to scale the detail level in all or part of the model, based on the particular simulation goal. In this manuscript, a proposal is formulated on how to structure a Modelica library so as to satisfy such a need.

DOI: 10.3182/20120215-3-AT-3016.00185

This work shows how the problem of modelling fluids motion can be addressed in Modelica. This innovative approach makes possible to face such a problem in a multi-physic modelling language as Modelica is. In this way it is possible to simulate together the fluid and the system that interact with it, without any additional effort and taking advantage of the Modelica libraries representing buildings, power plants, water treatment systems, HVAC and so forth.

DOI: http://dx.doi.org/10.3384/ecp12076477

Many studies for which simulation is necessary include the presence of control systems. While plenty of Modelica libraries are nowadays available to accurately represent the plant; the same is not so true as for the control elements; since industrial ones are endowed with a number of functionalities - and often present system or even vendor-specific peculiarities - that are not represented by the typical blocks (e.g.; based on transfer functions) offered by the existing libraries. This paper is an attempt to start filling the gap and provide an efficient solution; structured and organised in such a way to be easily understood by control specialists; and to ease information transfer between simulation studies and implementation.

DOI: 10-3182/20120328-3-IT-3014.00026

This paper presents a library of (PID) controller models adopting the object-oriented approach, and written in Modelica. Peculiar to this work is that controllers are represented both as dynamic continuous-time and digital models, achieving consistence between the two and accounting for the functionalities of typical industrial implementations. This allows the designer to use realistic controllers, maintaining the possibility of choosing the continuous-time or digital (event based) representation. The former allows for example for variable-step simulation, to the advantage of efficiency, while the latter represents very realistically the actual control system’s operation, clearly at the cost of more simulation time. Beside standard PI and PID controls, in this work also autotuning is (initially) considered, and some application examples are reported to show how the presented library can ease system studies involving (PID) controls.

The paper proposes storage tank models that are informative enough to study and assess the correct storage management - a relevant issue for energy efficiency - while being at the same time computationally efficient, and easy to integrate with models of different physical domains. The result is obtained by a convenient use of object-oriented modelling, via the Modelica language. After presenting the model, a simulation study is shown and discussed.

Simulation is important to evaluate the energy-related performance of a building, and for reliable results, reproducing the behaviour of the contained air volumes is particularly relevant. For such a purpose, fully mixed models (i.e., for instance, a single temperature per room) easily prove inadequate, while Computa tional Fluid Dynamics (CFD) ones are too complex, and difficult to formulate in a modular manner, to the detriment of their usefulness if the simulator has to be used throughout the project, and not only to assess its final result. This manuscript presents an intermediate solution based on the Modelica language.

Simulation is widely used to assess and/or improve the energy efficiency of both existing and new buildings. Such an analysis has to account for heterogeneous phenomena efficiently, to manage components in a modular manner, and (which is seldom addressed in a structured way) to scale the detail level in all or part of the model, based on the particular simulation goal. In this manuscript, a proposal is formulated on how to structure a Modelica library so as to satisfy such a need.

Autotuning methods are typically conceived as procedures, thus need simulating as digital blocks. However, when no autotuning is in progress, it is far more efficient to represent the tuned controller as a continuous-time system, to exploit variable-step integration. This manuscript presents the first nucleus of a Modelica library of autotuning controllers, where the problem just mentioned is tackled explicitly. The focus is here restricted to the PI structure, but the presented ideas are general.

DOI: 10.3384/ecp09430087

This paper focuses on the automatic generation of microcontroller code for hardware-in-the-loop simulation using Modelica models. In this work a test is presented and commented in which Modelica is used to specify a control system, the in-line integration code is obtained from the Modelica model and executed part on a PC, and part on a microcontroller board. The presented application, albeit created basically for educational purposes, covers quite different scenarii, therefore evidencing the usefulness of Modelica in the addressed context, and providing as a consequence some future research directions. The contribution aspect of this work is twofold: on one side the entire cross-compilator software chain is built within the same framework; on the other hand, all the involved software tools are open-source (mainly GPL) licensed, making the application extremely modular and extensible. Furthermore this work will be included in the next release of the open source Modelica environment SimForge, thus enriching its Modelica back-ends support.