S. Karatasou, Group of Building Environmental Research, Physics Department, University of Athens, Building PHYS-5, University Campus, 15784, Athens, Greece

M. Santamouris, Group of Building Environmental Research, Physics Department, University of Athens, Building PHYS-5, University Campus, 15784, Athens, Greece V. Geros, Group of Building Environmental Research, Physics Department, University of Athens, Building PHYS-5, University Campus, 15784, Athens, Greece

Abstract

Artificial neural networks (ANNs) have been used for the prediction of cooling loads of residential buildings in Athens, Greece. The investigation was performed for the summer period, where for Southern European countries, short time cooling load forecasting in residential buildings with lead times from 1 hour to 7 days can play a key role in the economic and energy efficient operation of cooling appliances.

The objective of this work is to produce a simulation algorithm, using ANNs, capable to forecast the following 24-hour cooling load profiles. Reliable cooling consumption measurements are required but are not usually available for residential buildings. State-of — the-art building simulation software, TRNSYS, was used to calculate energy demand for cooling for five selected apartments in Athens, Greece, using detailed building data (geometry, wall construction, occupancy etc) and Athens climate conditions. In order to validate simulation results and to have a more accurate modeling of difficult to estimate factors (like infiltration, absorptivity, inhabitants influence, etc), measurements of the indoor and ambient air temperature were performed during a period of two months, for April to May where neither heating nor cooling loads are demanding, and then used to calibrate the developed model of each house. Following this procedure, a database consisting of reliable cooling data was produced.

These data are used to train artificial neural networks in order to generate the relationship between selected inputs and the desired output, the next day building energy consumption for cooling. A multiplayer perceptron architecture using the standard back-propagation learning algorithm has been applied yielded to satisfactory results and the conclusion that when ANNs are trained on reliable data they can simulate the behavior of the building, thus they can be effectively used to predict future performance.

Introduction

In Southern European Countries, short term cooling load forecasting in residential buildings with lead times from 1 hour to 7 days can play a key role in the economic and energy efficient operation of cooling appliances.

In Greece, the operation of cooling has an important impact on the power demand profile. During the summer period, the highest peak power demand is approximately 2GW bigger than the bottom line peak, constant for the months April to May and October, when no cooling is in operation. The power profile between those months follows a similar pattern to the monthly average temperature profile. Furthermore, the fact that energy consumption for domestic use represents almost 35% of the total annual electricity consumption indicates that domestic load predictive models could be very useful from the Energy Utilities perspective.

To predict building energy consumption a large number of building software tools are available, making feasible to model a building for thermal evaluation and study it’s exact thermal behavior. Building thermal models which have been widely used in a variety of buildings and for a range of applications, in practice diversify on many factors: the modeling methodology, the physical laws, parameters and data that they encase, the integration of HVAC, passive solar, photovoltaic systems. Thus, depending on the application, these models vary on complexity and can be simple and easy to use, or more sophisticated and time-consuming to set-up and run.

In general, for the majority of applications, most of the appropriate software tools are time consuming and computationally heavy, especially when transient numerical methods are used. A large number of assumptions often need to be made when the quantitatively measurement of factors like infiltration or the estimation of parameters like occupancy is not possible. Also, parameters like the cost, the level of expertise and the exhaustive information needed to be collected could be prohibitive for a massive implementation.

Furthermore, almost all energy consumption predictive schemes are based on the prior prediction of weather data. As many weather variables are considered such as dry bulb temperature, relative humidity, solar radiation and cloudiness conditions, the most common practice is to use weather forecasts issued by meteorological centres, yet the direct link with such a centre make the procedure even more complicated. Artificial Neural Networks (ANN) can provide an alternative approach, as they are widely accepted as a very promising technology offering a new way to solve complex problems. ANNs ability in mapping complex non-linear relationships, have succeeded in several problems such as planning, control, analysis and design. They have been extensively used in the design, operation and fault detection of HVAC applications, in short time electric load forecasting, and in many fields of energy analysis and prediction.

In this study, a methodology based on a combination of an ANN and a thermal model is being used, in order to demonstrate the feasibility of using neural networks to forecast the following 24-hour cooling load profile for residential buildings. Neural network architectures are characterized by a collection of processing nodes connected by a series of weighted links. The relationship between the individual node’s inputs and outputs is typically a nonlinear function (for example a sigmoid function). A neural network can carry out complex calculations from global inputs to global outputs. In carrying out this process, and with the absence of a general theory for non-linear time — series prediction, good performance comes only by carefully analyzing available data (i. e. by using so-called ‘training data’).

Reliable cooling load data for at least one season are required but are not usually available for residential buildings, the focus of interest. Thus, dynamic building simulations were carried out with TRNSYS for five selected apartments in Athens, Greece, using known building data (geometry data, wall construction data, etc.) The simulation was carried out for a period of year, using as input for the climatic data the typical meteorological year. The training and forecast for the ANN model is then performed on the simulation results.

In the following we present the initial function requirements and the product finally developed.

1.1 : Function requirements.

a) Thermal insulation

The insulation concerns two aspects. Firstly the double glazing insulation has to be adapted to the thermal insulation of low energy houses with a positive energy exchange — out of collector function — over the year. Secondly the insulation of the solar collector has to limit the exchange with the inner of the building.

b) Solar collector

The solar collector function has to be optimized to insure part of the heating loads in the middle seasons and to reach a global productivity of 385 kWh/m2y (German criteria for subvention until June 2004).

c) Transparency

The initial target was to have a full visibility of 50% through the glazing. This point is in contradiction with the solar collector function.

d) Variable solar control

As a Venetian blind, dependant on the season and time, the aim is to favor the passive income and to limit the summer over-heating. This point takes into account the sun and daylight dependant on seasons and orientations too.

Over its technical functions, other criteria were also considered :

— Modular dimensions, as usual in conception of facades.

— Attractive design.

— Capacity to respect the normative tests of collectors and glazings.

Global target : The integration of the whole function in the glass collector has to be at least as good as each function in separate products. Further, our goal was to remain in conventional installation and connecting methods.

2.2: Composition.

The actual conception is a compromise between performances of the different functions. The diagram illustrates the composition of the glass collector composed of an insulated glass where a solar collector partially transparent is inserted inside.

Figure 1: Schematic cut of the glass collector The glass collector is composed of :

An insulated glass where the exterior glass, position 1-2, is an extra white glass. On the inside glass, position 3, is a vacuum sputtered low emittance coat. This coating principally limits radiant energy transfers between interior-exterior. Secondly, because of the low emissivity of the collector, this coating limits the transfers between the collector and the interior of the building.

The two glasses are separated by an edge spacer filled with desiccant to regulate humidity inside the double-glazing. The cohesion and gas-tightness is obtained by two glued barriers made of butyl and polyurethane. The interior of the double glazing is full of rare gas.

A solar collector is made of a copper serpentine where aluminum fins are “clipsed”. This fins are coated with a very effective solar-selective deposit with low emissivity and high absorption. A thermoplastic mechanical tightness guarantees the passage of the copper tubes crossing through the edge seal. In the serpentine, water collects and transfers the absorbed energy.

The reflectors are just behind the absorber, vacuum deposited directly on the glass, position 3. They tend to limit, like a Venetian blind, the solar gain. They reflect the intercepted solar radiation to the solar collector. This reflector deposit looks like a mirror on

both sides, with a design function too, because it hides the collector view from the inside as well.

So constituted, solar radiation can be separated in three operative parts figure 1:

• A first part, nearly constant, which directly intercepted by the solar collector.

• A second variable part, which is reflected from the reflectors behind the solar collector.

• A third variable part which is the passive solar gain.

With the sun rising, the amount of solar radiation intercepted from the reflectors increases in profit of the solar collector, until eventually to be total.

Adding equally sized obstruction blocks sidewise as shown in Figure 8, a typical atrium situation occurs. The factors stay unchanged. As Figure 9 indicates, the daylight amount is now reduced in average by 45%. An additional atrium roof glazing would leave an average 22% of daylight as compared to the base case (assuming an effective roof transmission of 40%). The horizontal obstruction size now shows a positive effect since the daylight open­ing increases with a.

Figure 9: Factors, definition bounds and resulting main effects on D for the case of the room daylit by one window with an atrium-like obstruction

When using structures with small periods, visual effects due to diffraction and refraction can be very distractive for the use in glazings, even if the overall function of the system in terms of total transmittance may not be strongly affected by miniaturisation of the structures. To study period dependent effects and to find an optimal feature size, we created two structures comparable to those in figure 1 with a period of 100 pm and a period of 17 pm. As shown below, there are slight differences of the structures due to the production process (see figure 4 and figure 8), but it is still possible to discuss the effects of the different feature sizes qualitatively. In figure 3, the measured relative intensity of white light as a function of the angle of transmission for
normal incidence is shown. For the large period (100 pm) , two narrow, high peaks were measured. One, at 0°, represents light directly transmitted through the coplanar areas and the other, at approx. -25°, light refracted by the prisms. For the use in a glazing, as a consequence this means that the user would see two different sharp images of comparable intensity at one position. For the small period the image based on refraction is strongly reduced in maximum intensity. It is also much more diffuse, because now there are several separated diffraction orders the intensity is divided into. The different position of this image is caused by slightly different prism angles. On the other hand, the first orders of diffraction are clearly separated from the central peak (zero order). For the use in a glazing, this means that the image generated by the refracted light can hardly be seen, but instead there will be clearly separated diffraction orders. This means that beside the main image one will see at least two colourful images close by, which are much less in intensity compared to the refracted image, but that are still distracting because they cause the main image to be blurred. There are several options to reduce these effects. The diffractive effects are most effectively reduced by an increase of the feature size. For refractive effects, one option is a face-selective absorbing coating, which decreases the intensity of the refracted image while the intensity of the direct image is unaffected. Tests showed that an intensity reduction of 50% already provides a good subjective visual improvement.

The tests began placing in the front wooden frame of the first module the clear glass, in the frame of the second one the filter glass, and on the third one the reflective glass. Eight type-T thermocouples were placed in the small scale test cell; two of them to measure the interior temperature, T, the temperature difference between the interior and exterior of the test roof, ATl, and the temperature difference between the interior and exterior of the glazing, ATv. Additionally, two thermocouples were placed to measure the ambient temperature, Ta. An acquisition system PCL-812G were used to record and process all the temperatures measured. To register the electricity consumption by the air conditioning systems, three sensors multifunction, form 2S, class 200 with integration time of 1 minute measured the amount of electricity consumed. These sensors storaged the measurements and at the end of each test, an optical cable was connected between the sensor and the acquisition system. The incident solar radiation impinging on window and on the roof was measured by using two Eppley pyranometers: a black and white pyranometer and a spectral precision pyranometer. Figure 3 shows the experimental set up during the testing. The wind velocity was measured with an ALNOR thermoanemometer.

The experimental test were performed at the city of Cuernavaca, Mbrelos, Mexico of 18.5o North latitude and 99 o longitude. The procedure was as follow: The three modules were placed outdoors facing south by 10 hours on July 7th, 2000. Temperature measurements were averaged each 30 seconds; solar radiation and wind velocity were measured and averaged every 10 minutes.

In works [6,7], on the grounds of thermodynamic description of non-equilibrium systems with the use of phase space formalism and the multi-dimensional fundamental equation at the availability of diffusion, the processes of chemical instability generating fluctuations in a macroscopic system including the vacuum cavity of a cryogenic reservoir with the superinsulation inside it have been explained. In addition, the macroscopic system consists of a large number of particles N ^ ■*> and occupies the macroscopic volume V^ ■*> at finite density N /V. Moreover, an external influence is directed to the system — in the form of low-magnitude fluctuations of the diffusive flow with the donor gas and water vapour concentration changing in composition. In work [6], small and intermediate magnitudes of fluctuations have been considered, which most effectively deflect the system from the unstable state.

This review covers the consideration and analysis of the data being accumulated by the present time on passing of electro-sorption processes in screen-vacuum heat insulation (superinsulation) layers of big cryogenic reservoirs.

Periodic variations of the concentration of the effusion flow in the link with electro-sorption processes lead to the appearance of chemical pseudowaves. In a diffusion system with chemical reactions, information is transmitted at an infinitely high speed, as such a system related to the parabolic type. Therefore no delay periods are observed between the rate of change of the concentration parameters of effusion values and the variation of thermodynamic parameters in the thermodynamic open macrosystem being investigated. The influence of chemical pseudowaves on the
cryogen product volatility as well as on reduction of the safety degree of thermally controlled objects have been analysed.

At the investigation of a non-adequate process, the variations of volatility in identical cryogenic reservoirs during purposeful reduction of the hydrogen concentration in a clearly expressed hydrogen residual atmosphere the following effects have been detected [6,7]:

1. Effect of effusion induced hydrogen instability of the superinsulation (EIHIS)

[6,7],

2. Effect of effusion induced heat conduction instability of the superinsulation in cryogenic and vacuum facilities (EIHCIS) [6, 7],

3. Effect of multiplication of the amount of desorbed hydrogen molecules in respect to the magnitude of inflowing humid air molecules in the superinsulation of cryo-vacuum objects (MADHM) [6, 7].

The effects in heat insulation can be controlled. In order to create new heat insulation samples with a high exergy efficiency and a high safety degree, new heat-insulating structures and designs should be developed [8, 9].

As a rule, in the process of operation of big cryovacuum objects, when the heat insulation routine maintenance intervals are exceeded, a process of daily fluctuations of the residual pressure and volatility of the cryogenic liquid arises. The cryogenic reservoirs in question of the RS-1400/1.0 type (hydrogen, nitrogen, oxygen) [10] had an insignificant atmospheric effusion leak being within the tolerance by magnitude. Variations of the residual pressure and volatility of the cryogenic liquid occur in such heat-insulating cavities (HIC) with expressed symbate nature in respect to the variations of the atmospheric leak effusion component.

However for the residual pressure variations, the multiplication mode of desorption processes is characteristic as compared with the calculated effusion flow. The influence on the residual HIC atmosphere by a selective chemical hydrogen absorber based on palladinised manganese dioxide [30,31,34] has allowed to reduce the cryogenic liquid volatility from 1100 kg/day to 390 kg/day [6]. The oscillatory process of the volatility has stopped. The oscillatory process of the residual pressure occurred with the monotonously decreasing correlation coefficient (the variation of the residual medium pressure — the relative humidity) as the hydrogen was pumped out from its maximal value of 0.95 to negative values. After pumping out of the calculated amount of residual hydrogen, the correlation coefficient has changed its sign and has been monotonously increased by modulus up to the value of

0. 95. The process of reduction of the correlation coefficient has been symbate in respect to the process of hydrogen removal from HIC. The change of sign of the correlation coefficient occurred at the moment when the whole calculated amount of hydrogen had been removed from HIC [6]. Such behaviour of the correlation coefficient has caused an idea of the electronically stimulated adsorption-desorption process of acceptor or donor gas depending upon the Fermi level of the metallised screen surface. Two hypotheses have been considered:

1. Adsorption-desorption process thermostimulated in micropores.

2. Electronically stimulated adsorption-desorption process of acceptor or donor gas (depending upon the Fermi level) by the inflow of ionised oxygen and water vapour.

The Renewable Energy Centre is located in a relatively affluent part of the United Kingdom. However, the relocation of an expanding company to Kings Langley will provide opportunities for work and provide alternative career possibilities outside the magnet of London, obviating the need to commute. The new facilities will assist RES in expanding their operations worldwide and the creation of wealth inherent in this expansion.

Transference of the content and the experience gained from the design, installation, commissioning and testing of the energy systems to those in the construction, renewable and property industries and businesses in the UK will foster the growth and development of them. The professional and commercial exploitation of design strategies, installation design and new products resulting from the project will also assist development of them.

Life Chances

The main social benefit locally will be the provision of an efficient and stimulating workplace. However, the decision to operate the new head office as a visitors’ centre and information resource, allowing those of all levels of interest to learn about the technologies and issues involved in creating low and zero net energy work settings, provides an invaluable national facility. The web site allows access to this data and information world

Clean and Green

Bringing back to life a derelict building rather than building new is a considerable benefit in terms of land utilisation, use of resources and improving the amenity of the area. The construction work was undertaken on the basis of minimising waste and using materials and components with low embodied energy from readily available resources.

In order to minimise the need for energy, a judicious combination of active systems (mechanical ventilation, artificial cooling, heating and lighting, building management systems) and passive systems (solar heating, natural ventilation and lighting, solar shading, a well insulated building envelope incorporating thermal mass) was developed.

The buildings are exposed to considerable external noise from passing trains to the west and the motorway to the south. To cut out the disturbance from noise inside the buildings, the outward facing facades had to be sealed. This, together with the relatively high levels of heat generated by modern office use, requires the building to be artificially cooled in summer months. The cooling source is water drawn from aquifers located in the chalk below the building. This strategy avoids the heavy energy consumption and potential polluting effects of refrigeration plant normally used for air conditioning. The cool water is used to drop the temperature of air being fed into the building and/or is circulated through convectors within the office space, cooling the air within it.

Heat is supplied from the biomass boiler (or gas boiler until such time the biomass plant is installed) and from the PVT array, either direct or via the seasonal ground heat store. Hot water from these sources is used in a similar way as the chilled water for cooling. Electricity is generated from the PVT array and the wind turbine.

Windows can be opened in facades and roofs facing away, or sheltered from, the motorway and the railway, to ventilate the building in temperate conditions. Exposed
windows are shaded from the sun by fixed glass or aluminium screens and by deciduous tree planting, thereby reducing unwanted solar gains and the need for cooling. The building is well insulated and sealed.

Predicted energy use and energy supply is shown in the table below. The current monitoring programme will show whether these predictions are born out in reality.

Estimated energy balance for the site:

A building management system (BMS) controls and optimises all the energy systems, including opening and closing the roof lights. It also records all monitored results from the various energy systems before passing the results to a site in Denmark for uploading onto the website.

RES actively encourages staff to use public transport, bicycles and car sharing for travel between home and office.

About 5ha of the 7.5ha site are given over to miscanthus cultivation. In addition there is a car park and a 5 aside football pitch. The remainder of the land is planted with indigenous species of trees, shrubs and grasses. Wild life is encouraged by the re-creation of natural habitats.

The test building is located in the outskirts of Milano. At the first floor there are two test rooms having the same volume, the same exposure of the windowed facade and the same internal loads. The rooms are equipped with radiant panels formed of small pipes, installed into the ceiling in test room 1 and into the walls in test room 2. The pipes are made of polypropylene and have an external diameter of 3.35 mm and a thickness of 0.5 mm.

The experimental system layout is shown in Figure 1. It consists of two water loops: a rooms loop, containing the radiant panels and removing heat from the building, and a ground loop, containing the earth-to-water heat exchangers and discharging heat into the ground. Thermal interaction between the loops is achieved through a counter current flow heat exchanger. Each loop is provided with a pump.

The ground heat exchanger consists of 10 vertical steel pipes 6 m deep connected in parallel. The pipes are arranged in two parallel lines in a rectangular pattern. The distance between adjacent tubes is 1.5 m. Each element consists of two concentric tubes so that warm water coming from the building flows down through the hollow space between the outer and the inner tube and flows back into the inner tube. A layer of insulating material
covering the internal tube prevents heat transfer between the descending and the ascending fluid.

A monitoring system collects data with a time step of 10 minutes. The following quantities are measured:

— in the test rooms: air temperature, mean radiant temperature, radiant panels surface temperature, air humidity

— in the rooms loop: water temperatures at the inlet and the outlet of the radiant panels

— in the ground loop: water temperatures at the inlet and the outlet of the earth-to-water heat exchanger

— in the ground: temperatures at different depths (1.5, 3, 4.5 and 6 m) in the perturbed and in the unperturbed zones

— meteorological quantities, i. e. outdoor air temperature and humidity, wind speed, global horizontal solar radiation

The water mass flow rates are measured with variable area flow-meters by manual reading. Power consumption of the pumps is measured through electricity meters.

The greenhouse structure includes two linear, three-storey wings that are arranged in two rows along a central axis. Inner spaces under glass are landscaped with water and greenery, creating a large-scaling winter garden. In the winter it works in tandem with the concrete and gravel floors to collect solar energy, while acting as a thermal buffer-zone. In summer PV modules integrated with glass roof act as shade-system device protecting against overheating and from too much light. The parts of the fagade can be opened to ventilate the greenhouse through natural cross ventilation.

ARCHITECTURAL CONSEQUENCES.

The architecture of the building may be defined by selecting its features that include: urban matters (especially building’s closest surroundings), its function with sort of utility process in the interior, structure and aesthetics.

The real and potential impact of PV modules’ usage on the inner space environment may have its response in some of these features, causing architectural consequences.

Three different types of glazings with various U-values and transmittances are studied in this paper: one "standard” clear double-glazed window (clear DG), one low-e coated double-glazed window (Low-e DG) and one low-e plus AR-coated double-glazed window (AR Low-e DG), Table 1.

The energy simulation tool ParaSol v2 was used to simulate the monthly average direct and total solar energy transmittance (Ts0| and g-value) as well as the annual energy demand. The solar transmittance, Tsol is the transmittance of the glazing for the entire solar spectrum. ParaSol simulates the monthly average value of Tsol, taking into consideration the actual climate and solar angles. The g-value (total solar energy transmittance) is the solar transmittance plus the absorbed heat in the window panes emitted to the inside. Consideration of potential overheating problems was done by looking at the number of overheating hours. ParaSol simulates a room module with only one wall and one window that abuts to the outside climate. The other three walls, floor and ceiling abuts to other rooms with the same indoor temperature i. e. adiabatic walls. ParaSol is a freeware developed by The Division of Energy and Building Design, Lund University (Bulow-Hube H. and Wall M). It is available at (http://www. parasol. se).

Each glazing type was studied in three different Nordic climates: Copenhagen (DK), Stockholm (SWE) and Helsinki (FIN). The Parasol simulations were also done for all 4 major directions (N, E, S, W) and the average was then taken from these four simulations. The ParaSol simulations were used to evaluate the potential of energy savings by using AR — coating on a low-e DG.

The room size investigated was 20 m2 (L x W x H = 5.0 m x 4.0 m x 2.7 m) and can be regarded as a typical living room. The temperature set point was 20°C for heating. No consideration of the cooling demand was done in this paper, since air-conditioning is not common in residential buildings in Scandinavia. The ventilation rate was set to a constant value of 0.6 ach. The internal heat load was 5 W/m2 both day and night. The glass area was assumed to be 70% of the window area, or approximately 2.1 m2. See Table 2.

Table 2 Input data used in the simulations of the room.

Exterior wall U-value

Room size L x W x H

Window measurement incl. frame

Clear DG U-value excl. frame / incl. frame

Low-e DG U-value excl. frame / incl. frame

AR Low-e DG U-value excl. frame / incl. frame

Ventilation (00-24)

Internal heating load (00-24)

To simulate the daylight availability we used Rayfront v1.04. Rayfront is a user interface to the lighting simulation software Radiance which is the industry standard raytracing engine for physically correct lighting simulations.

The Rayfront simulations were performed to obtain the daylight factor, which is a measure of the ratio between the interior illuminance and the exterior illuminance from an unobstructed overcast sky, see equation 1. The daylight factor was calculated 0.8 m above the floor level. Since the daylight factor is independent of orientation and latitude, it was only studied for one location (Stockholm) and one orientation. The simulations were made for the standard CIE overcast sky with the reference illuminance value of 13826.73 lux (default value in Rayfront) for the 21st of June.

The daylight factor is defined as:

Ei

DF = *100% Eq.1

Eo

Ei= daylight illuminance on indoor working plane

Eo= simultaneous outdoor daylight illuminance on a horizontal plane from an obstructed hemisphere of overcast sky

In Rayfront the transmittance of the window is given by the transmissivity parameter. The transmissivity is calculated from the light transmittance of the window according to the following formulas:

The total light transmittance of the window with two panes (index 1 and 2) was obtained from:

1 — Rsi* Ris2

Where index 1 and 2 refer to the transmittance T and reflectance R of the two panes. The transmissivity was then calculated from the light transmittance in Eq.2:

vis

Equation 3 was found in the Radiance manual. The equation recalculates the light transmittance to transmissivity because Radiance uses the transmissivity parameter instead of the transmittance.

The properties of the different panes with and without coatings are described in Table 3.