Towards Zero Energy Balance in TertiaryBuildings ?

A. Lluna I. Benıtez J. Monreal I. Dıaz

Energy Technological Institute, Paterna, Valencia, 46980 Spain (e-mail:andres.lluna@ite.es,ignacio.benitez@ite.es,javier.monreal@ite.es,inma.diaz@ite.es)

Abstract: The present work details the development of an integrating system of energy pro-cesses and resources within tertiary buildings, including the design of an intelligent managementsystem that learns and predicts future scenarios and actions based on historical data. Thework being done includes the development of a centralized data warehousing system, energyconsumption and generation monitoring and visualization, and full SCADA and automationdesigns and systems integration, concerning renewable energy resources, energy consumptionsand the addition of a solar energy based refrigeration system. Above all the different levels, anintelligent control development produces the adequate actions to manage all the processes basedon dynamic models and prediction of scenarios and future outcomes.

Keywords: Automatic process control, Distribution automation, Communication protocols,Energy management systems, Energy monitoring, Modelling, Simulation, Intelligent control.

1. INTRODUCTION

The Energy Technological Institute (ITE) has as one ofits objectives the improvement of energy efficiency at ter-tiary buildings, aiming to achieve a reduction in energyconsumption that may lead to a zero energy balance. Inthis context, the following paper describes the work beingdeveloped to integrate all the energy subsystems of anoffices’ building (energy consumption, distributed gener-ation, energy storage, weather conditions) under a global,intelligent management system, able to gather informationfrom the different processes, and apply machine learning(Nilsson (1996); Bishop (2006)) and intelligent controltechniques (Yu (2009)) in order to produce the necessaryactions to achieve an improvement in energy consumption.

1.1 Scenario of analysis and development

The development of the systems integration and controldesign is being applied on the ITE offices building. Thereare a number of existing resources and infrastructures inthe building, dealing with renewable energy resources andenergy integration. These are:

• A pilot plant that implements generation from threedifferent resources: photovoltaic (PV) panels, windturbine and a PEM fuel cell.

• Energy storage, in the form of batteries of differenttechnology, and supercapacitors.

• A solar-energy fired absorption machine refrigerationsystem, which is being implemented to aid the ex-isting conventional heat pump of the building. This

? The development of this work has been supported by the IMPIVA(Institute of the Small and Medium-sized Industry), which is at-tached to the Conselleria de Industria, Comercio e Innovacion (In-dustry, Trade and Innovation Regional Minister) of the Valencia Re-gional Government, and the European Regional Development Fund(ERDF) from the European Commission.

system includes a deployment of conveniently sizedsolar collectors, in order to provide hot water fromsolar energy to the absorption machine.

Within this context, the global objective is focused onstudies, research and development aiming to optimize en-ergy consumption, a fully integration of systems and man-agement and tools to easily gather, visualize and recordhistorical data concerning all the relevant variables ofenergy consumption, generation, and climate conditions.This project brings into a common specification differentresearch areas related with energy efficiency, optimiza-tion and distributed renewable energy resources, such asdomotics and comfort conditions, process control, SmartHomes (Augusto and Nugent (2006)), machine learningand artificial intelligence (Himanen (2003)), and systemsengineering.

1.2 Objectives

This work follows the ITE concern with sustainability, en-ergy efficiency, systems integration and renewable energyresources, either in tertiary or residential buildings. Spe-cific objectives of this implementation are the following:

• Improving the energy balance of a building by opti-mization of energy resources, both conventional andrenewable, and storage options, by improving the useof these resources, their operation, management andinteraction.

• Optimal integration of all the energy systems (gener-ation and demand) within a global framework, ableto monitor all the variables and produce controlactions or automate reference values generation byimplementing decision algorithms based on machinelearning and artificial intelligence techniques.

• Develop a centralized data warehousing system, toacquire, process, store and visualize data from all theprocesses and systems being integrated.

• Develop and integrate data mining techniques (Jack-son (2002); Han and Kamber (2001)), able to describepatterns of use and discover interesting relationsamong generation, demand and climate conditions,and to build prediction models able to anticipatefuture outcomes and energy needs.

• Implementation of advanced communication proto-cols and networks to allow efficient data acquisitionand systems control of all the processes involved.

• Development of the appropriate automation andSCADA systems to integrate and control all the pro-cesses involved in the building’s energy systems.

• Efficient integration and management of distributedrenewable energy resources, energy storage devices,data acquisition and systems control within a cen-tralized management application.

• Development of dynamic models to define the be-haviour of all the components and subsystems in-cluded in the design. Simulation of scenarios and de-sign of advanced control techniques to improve energyefficiency and systems integration.

• Distributed monitoring and visualization of all theenergy consumptions, generation and environmentalconditions at the building.

• Implementation of domotics or home automationsystems for lighting and refrigeration management.

• Integration of solar energy with the existing refriger-ation installation at the building.

• Development of autonomous decision-making sys-tems, able to analyze all the available information andmake decisions that anticipate the future scenariosof energy consumption and demand, based on intelli-gent control techniques and the multiagent designingmethodology (Ferber (1999)).

These tasks are being currently developed, grouped underthree research and development areas. At a first stage,a task on modeling and simulation has been performed,with the objective to obtain dynamic models of the re-newable energy resources and the solar energy refrigerationplant, in order to simulate different scenarios and providevaluable information that will help to design advancedintegration and control systems.

Following, an automation task deals with the physicalimplementation of the following aspects:

• A distributed control system of the solar energyrefrigeration plant, and its connection and interactionwith the existing heat pump of the building.

• Automation and process control of the refrigerationsystem of the building.

• Monitoring and visualization of all the energy vari-ables considered, such as: electric energy used, asregistered by the building’s meter; energy generationfrom all the available energy resources; energy con-sumption in refrigeration measured in specific meet-ing rooms of the building; generated solar energy fromthe solar collectors to feed the absorption machinesystem.

• Data warehousing of all the monitored data, includingredundancy storage and easy accessibility.

• Integration of the mentioned systems, by design andimplementation of communication protocols, central-ized applications and visualization options.

Finally, a higher level of management is being studied,based on artificial intelligence, with the objective to designan intelligent control system, able to produce its own deci-sions in order to improve energy efficiency and integration,assuring the comfort margin.

These steps of design and development are described inthe following sections.

2. SYSTEMS MODELLING AND SIMULATION

The objective of modelling and simulation is focused onthe following issues:

• Modelling and simulation of the available renewableenergy resources (PV panels, wind turbine and PEMfuel cell).

• Modelling and simulation of the refrigeration systembased on solar energy and absorption machine.

• Modelling the systems integration of building’s heatpump and solar energy refrigeration system.

The two first items are described next, whilst the third isstill in its initial stage.

2.1 Modeling and simulation of renewable energy resources

In the context of the integration of distributed renewableenergies, the ITE has set an experimental plant for electri-cal energy production and supply, which implements PVmodules with a total of 7.5 kWe (2.5 kWe per phase), awind turbine of 6 kWe, a 4 kWe PEM Fuel cell, a 6 kWeElectrolyzer, and a meteorological station, that collects thevalues of different weather variables. A research has beendone to analyze the relation among the different combi-nations of weather conditions and the resulting outcomeof electrical energy production from the PV modules andwind turbine.

The methodology applied makes use of multilayer Arti-ficial Neural Networks (ANN) (Haykin (1994)) to buildprediction models of electrical energy production, as afunction of the climate conditions. The input and outputvariables chosen for the models have been selected basedon a correlation analysis. The ANNs have been trainedand validated making use of monitored data of produc-tion and weather conditions, through the years 2007 and2008. Figure 1 displays, for instance, predicted values ofactive power generated by the PV panels, as a function ofambient temperature and solar irradiance.

The PEM fuel cell has been modeled making use ofrecursive ARMAX models, with the structure written in1. The data to train and validate the model has beenextracted from different experiments made through theyear 2008, varying initial conditions and load values.Figure 2 displays the training process of a model of powergeneration from the cell.

A(q)y(t) =nu∑i=1

Bi(q)ui(t− nki) + C(q)e(t) (1)

Fig. 1. Active power prediction from the ANN for thephotovoltaic panels.

Fig. 2. Modelling PEM fuel cell based on experimentaldata.

2.2 Modeling and simulation of the solar energy refrigerationplant

A simplified schema of the system is depicted in Fig. 3. Thesolar energy system has been designed to provide either hotor chilled water, according to the seasonal or climate needs.This is accomplished by switching the flow of hot waterfrom the solar collectors to a heat exchanger, whenever hotwater is needed, or to an absorption machine, wheneverchilled water is demanded. The resulting energy transferis directed to the building’s heat pump. Other workingmodes have been designed, concerning the load and supplyof hot water from two storage tanks, for an extra aid of hotwater, and an auxiliary heater, provided the heat from thetanks and the solar collectors are not sufficient to reach thedesired temperature (e.g., at night or late in the evening).All the different operation modes are controlled by three-way ON/OFF valves. The design is completed with acooling tower, as required by the absorption machine toreject the excess heat from the absorber and the condenser.Flow is controlled by variable-speed water pumps, andtemperature can be regulated by two proportional three-way valves, one at the entrance of the solar collectors area,and another at the entrance of the cooling tower.

The software DYMOLA has been used to dynamicallytackle the behavior of the plant, taking into account the

Fig. 3. Simplified diagram of the solar energy refrigerationplant.

Fig. 4. Model of refrigeration plant - heat production fromthe solar collectors.

different working modes, the changing weather conditionswhich will affect solar energy production and, therefore,the transient effects that will take place when switchingmodes or regulating water pumps and valves. An entiremodel of the plant is being developed, making use ofavailable libraries dealing with fluid mechanics and ther-modynamics, such as the Buildings (Wetter (2009)) or theThermoPower (Casella and Leva (2009)) libraries.

Figure 4 depicts the model of one of the available workingmodes, heat production directly from the solar collectors.DYMOLA allows simulating and visualizing all the definedphysical variables of the model and its elements. Forinstance, Fig. 5 depicts dynamic heat transfer processtaking place at the heat exchanger.

Each working mode is being modeled as a standaloneplant. However the development process will include afurther development where the plant will be modeled asa hybrid system, with continuous states or working modesand regulation (flow, temperature), being switched bydiscrete events (the three way valves that make the plantswitch from one working mode to another).

3. SYSTEMS INTEGRATION AND AUTOMATION

3.1 Distributed systems integration

The ITE building presents a number of distributed subsys-tems, providing diverse functionalities. These subsystemscan be differentiated in:

• Those that make use of energy: lighting system,HVAC system, electric circuit sockets, data process-

Fig. 5. Simulation of dynamic temperature evolution atthe heat exchanger.

ing center, kitchen, laboratories and operating ma-chines, and other facilities within the building.

• Power generation systems: wind system, PV system.• Energy storage systems: batteries, super capacitors,

hydrogen fuel cell.• Support systems based on renewable energy re-

sources: thermal plant.• Control and monitoring systems: each system de-

scribed has its automatic control system or monitor-ing system, based (mostly) in a development done bythe ITE. Generally, this control is performed usingindustrial control and automation techniques, keepingin mind the appropriate control for the application.

3.2 Light management

For lighting control of certain rooms of the building, aKNX protocol-based home automation system has beenused, that is being integrated into the overall architectureof the building. The KNX is a standard, defined bythe KNX Association (www.knx.org), for communicationsystems in domotic applications and control. The resultsof these improvements will be compared to the rooms thatdo not have this kind of technology.

3.3 HVAC Control

The HVAC or heat pump system of the building is cur-rently being optimized, by implementing developed Pro-grammable Logic Controller (PLC) automation with PIDcontrol of each zone of the building. This system willbe properly monitored, and controlled into the overallmanagement of the building.

3.4 Solar energy refrigeration plant control

The control of the refrigeration plant and the absorptionsystem has been made with PLC technology. To governthe necessary actuators and to acquire information of the

Fig. 6. Automation and SCADA interphase of the solarenergy refrigeration system.

sensors used, due to the decentralization of the elementsalong the process, an industrial control bus (DeviceNet)has been implemented, which offers versatility, robustnessand simplicity in the integration of elements. This systemis monitored, supervised, and controlled from the central-ized management system of the building. Figure 6 displaysthe supervisor interphase of the control system.

3.5 Renewable energy pilot plant control

The pilot plant includes the renewable energy resourceswind and solar PV system, in addition to a hydrogenPEM fuel cell, storage systems, and test system. Allthese systems are monitored by PLC technology and theinformation is integrated into the building managementsystem. In future steps a remote control and managementsystem will be implemented for this plant.

3.6 Description of the monitoring architecture of the totalenergy consumption and generation of the building

Due to the characteristics of the application implementedand the functions of the monitoring system (distribution ofenergy data collection in many labs, electrical panels at dif-ferent locations of the building), a distributed architectureof acquisition has been implemented, with a centralizedcomputerized system of data storage and management.The conceptual model proposed for this monitoring systemis depicted in Fig. 7.

Four hierarchy levels are proposed, which describe thesystem starting from physical, field levels, to the communi-cation and information channels, and to higher abstractionlevels dealing with the storage, displaying, processing,analysis and management of the acquired information,properly stored and centralized.

At a field level (Plant-floor level), energy meters have beenchosen as a key element to capture the energy state ofthe building. These elements have been chosen followinga criterion of heterogeneity in brands and communicationsystems to integrate them into the global system. There-fore, the maximum number of technological possibilitieshas been integrated, demonstrating the flexibility and scal-ability of the design. These devices are distributed through

Fig. 7. Architecture levels of implementation.

Fig. 8. Monitoring electric current consumption of thebuilding.

the building, measuring energy consumption at strategicpoints. Figure 8 depicts, for instance, the monitoring ofelectric current consumption of the building, in workingand non-working days.

These measurement devices are centralized by dispersionof location, through the Distributed Control and DataAcquisition systems. These devices are PLCs, which col-lect information from networked sensors, processing andstoring the data in order to transfer it to higher lev-els. PLCs provide scalability and modularity, robustness,programming capabilities, storage capacity, and differentcommunication ports, and field bus implementations.

In the Communication Level the PLCs transfer data to thelevel of information through a local area network availablein the Institute. The interface is a part of the System LevelCentralization and Information Storage, that acquires, ata desired frequency, the data, storing the informationcoherently in databases. Another component part of thislevel is the visualization interface.

Finally, in the Analysis Level the data are treated and an-alyzed using advanced techniques of data analysis in orderto automate the conclusions to be obtained: reporting ofabnormal behavior, system failures, energy optimization,consumption patterns, etc. These tools will allow knowingthe energy state of the building and possible actions to betaken. Figure 9 displays the integration of the monitoringsubsystems at the building.

In the next step and with all the information processed,intelligent control techniques are proposed to be devel-

Fig. 9. Global communications integration.

Fig. 10. Fuzzy inference to manage energy resources inte-gration.

oped, that will provide autonomous decision for the controlsystems aiming at reducing energy consumption of thebuilding.

4. INTELLIGENT CONTROL DESIGN ANDCENTRALIZED MANAGEMENT

The prediction models of renewable energy resources allowfurther studies on intelligent hybridization techniques, as afunction of the building’s needs and the climate conditions.Concerning the pilot plant , a research is being done onthe development of renewable energy sources management,based on fuzzy sets (Zadeh (1965)). The objective is toautomate the use of the different energy resources (PV,wind turbine, fuel cell, batteries), according to the climateand the prediction of energy consumption at the building.An example of this design can be seen in Fig. 10.

Concerning the refrigeration plant based on solar energy,the dynamic models of the plant built with DYMOLA

will allow a further design in process control, with theobjective to improve the transient behaviour and achieve abetter energy performance. In this sense, advanced controltechniques for hybrid systems, such as model predictivecontrol (Alessio and Bemporad (2007)) are being studied.

Upon these control levels, a global, centralized manage-ment system is being studied, based on intelligent controltechniques, that will make use of all the available informa-tion and measures, and the obtained models, to provideautomated decisions that will anticipate the building’senergy demand.

5. CONCLUSIONS AND FUTURE WORK

This paper has presented the overall work and researchbeing developed with the objective of improving the build-ings’ energy efficiency and in particular the ITE officesbuilding. This improvement is based on a correct useand balancing of resources, not only consumption, butalso generation and storage resources. In this sense, theresearch line started by the ITE aims to zero energybalance concepts in tertiary buildings and by extensionhome buildings.

The approaches to achieve this concept are: a preciseknowledge of the building’s energy state (distributed mon-itoring system of energy variables, functional control sys-tem and boundary and environmental conditions); a cor-rect and optimal integration of all the functional and op-erational subsystems of the building (management systemthat unifies all the subsystems of consumption, generationand storage); the mathematical modeling of componentsand simulation of subsystems and environment, along withimplementation of data processing techniques to studythem; and the development of intelligent control tech-niques for designing of automated decision systems, ableto manage the total resources of the building.

The development requires the integration of many hetero-geneous and functional systems distributed throughout thebuilding. Each of these systems presents its own featuresfor communication and use. This fact poses a challengein achieving a true interoperability among them. In thissense, it is important to highlight the difficulty of thistask due to the huge number of technical solutions thatare commercially available, each one of them with theirown functional and communication specifications.

The current challenge is to understand the house orbuilding as a ”whole”, controlling the entire environmenttogether as an entity. This control approach of a buildinghas proven to be extremely complex because there areno valid models due the heterogeneity of the systemsconcerned.

This development proposed is contextualized and cus-tomized for a particular context, the building of ITE,where a number of particular systems of consumption andgeneration are found. However, the proposed methodologyand techniques used can be translated to other contexts,taking into account the particularities and the basis ofan empirical knowledge of the application site. From theperspective of scalability of the system, it is importantto emphasize that is appropriate and desirable to formal-ize centralized management of buildings, considering even

districts management, in order to obtain an overall resultof improved energy. The proposed methodology allows toimplement these desired objectives to higher hierarchicallevels.

As future lines of this work the following are proposed:

• Advanced control design for refrigeration, either pre-dictive, fuzzy, or other advanced control techniques.

• Application of intelligent control for the HVAC ofthe building in terms of input parameters: climate,activity in the building, needs, etc.

• To design and develop a distributed network basedon the multi-agent approach (Cook et al. (2006)),integrated by sensors and actuators that interact witheach other without any centralized control. The au-tonomous behaviour will help in the achievement ofcomplex goals in such heterogeneous systems, provid-ing them with a capacity for very high performanceand interoperation.

• Finally, a design and implementation of a global con-trol system is proposed, with capacity to take deci-sions autonomously, with a high level of energy man-agement, and decisions closed to human abstractionlevels, aiming to obtain energy sustainability near thezero balance in the building, without affecting thecomfort limits.

REFERENCES

Alessio, A. and Bemporad, A. (2007). Decentralized modelpredictive control of constrained linear systems. 2813–2818. Kos, Greece.

Augusto, J. and Nugent, C. (2006). Smart homes can besmarter.

Bishop, C.M. (2006). Pattern Recognition and MachineLearning. Springer.

Casella, F. and Leva, A. (2009). Thermopower library -modelica library for the dynamic modeling of thermalpower plants.

Cook, D., Youngblood, M., and Das, S. (2006). A multi-agent approach to controlling a smart environment.

Ferber, J. (1999). Multi-Agent Systems – An Introductionto Distributed Artificial Intelligence. Addison Wesley.

Han, J. and Kamber, M. (2001). Data Mining: Conceptsand Techniques. Morgan Kaufmann Publishers, SanFrancisco.

Haykin, S. (1994). Neural Networks, a ComprehensiveFoundation. MacMillan.

Himanen, M. (2003). The Intelligence of Intelligent Build-ings. The Feasibility of the Intelligent Building Conceptin Office Buildings. Ph.D. thesis, Helsinki University ofTechnology.

Jackson, J. (2002). Data mining: A conceptual overview.Communications of the Association for InformationSystems, 8, 267–296.

Nilsson, N.J. (1996). Introduction to machine learning.Textbook Draft, Stanford University.

Wetter, M. (2009). Modelica library for building hvac andcontrol systems r&d.

Yu, W. (ed.) (2009). Recent Advances in IntelligentControl Systems. Springer.

Zadeh, L. (1965). Fuzzy sets. Journal of Information andControl, 8, 338–353.