How to Simulate Thermal and Fluid Dynamical Processes in Closed

How to Simulate Thermal and Fluid Dynamical Processes in Closed
Greenhouses including Water Interactions between Plants and Air
P. Jochum and M. Buchholz
Technical University of Berlin, Faculty of Architecture
Straße des 17. Juni 135, 10623 Berlin
Keywords: CFD, closed greenhouse, natural ventilation, simulation, Smile, thermal and
pressure node model, water treatment
A new type of greenhouse is presented. It is a closed system using buoyancy as
driving force for ventilation. The greenhouse is equipped with a tower in which the air
is cooled or heated to maintain the inner temperatures in the required band and then
conducted back to the plants. During the cooling process the water in the humid air
condenses and can be reused. The subject of this article is the way of prediction of the
thermal behaviour of closed greenhouses by means of dynamic simulations. Several
model approaches are discussed. The simulation environment Smile is presented as a
flexible program for this purpose.
Closed greenhouses were designed hoping for various advantages. First the possibility of injecting CO2 without important losses for increased fruit production, second the
reduction of water losses by the absence of wind and the potential of reusing water from
condensation processes, and third the use of the greenhouse as a solar collector extracting
excess energy during hot days for heating purposes in cold periods thus stretching its
operation time without external energy supply. The main disadvantage of closed
greenhouses in contrast is the non trivial thermal and humidity behaviour of the combined
system consisting of greenhouse cover, air, plants, soil, and heating/cooling facilities.
Therefore an EU funded research project, named Watergy, has been created
(Buchholz et al., 2005). In this paper the chosen approach to the simulation of the
investigated systems will be shown.
The objective of Watergy is the development of a closed humid air solar collector
(Fig. 1) driven by natural convection forces due to buoyancy. The solar collector could be
either a solitary greenhouse or an attached greenhouse in front of a building. The
buoyancy resulting from the warmed up plant region causes an air flow to the top of the
collector. From there the air will be conducted back to the plants through a colder heat
exchanger which is installed in a return duct. The cooled air flows back to the plant region
where the cycle restarts. During the circulation of the air the moisture content varies in a
wide range. The air will be humidified by the transpiration of the plants and the
evaporation from the soil. Furthermore the air is additionally humidified with grey or
salty water. To prevent contact of this water with the plants, an extra so called inner roof
has been installed collecting the remaining drain water that is not evaporated. In the cold
heat exchanger the water vapour will condense and can be collected for irrigation or other
purposes (Fig. 2).
Through the humidification of the air by evapo-transpiration and the additional
sprinkler system the air temperature can be kept relatively cold. This results in smaller
heat losses to the environment. A heat storage is needed to extract the heat from the
collector and supply the heat exchanger with cold water. Cold water can be obtained by
heating the greenhouse with the same heat exchanger during colder periods, in the case of
diurnal storage during the night or in the case of a seasonal storage during the colder
Proc. IC on Greensys
Eds.: G. van Straten et al.
Acta Hort. 691, ISHS 2005
To acquire realistic results regarding the behaviour of the described humid air
collector, the system is analytically and experimentally studied (Buchholz et al., 2005).
For the experimental investigations two prototypes will be erected. The first prototype, a
greenhouse with diurnal storage, has been built in the south of Spain (Almeria),
measurements (Janssen et al., 2005) starting in July 2004. The second prototype is an
office building connected to a greenhouse with a seasonal storage. Constructions are
going to start in March 2005.
The idea of creating free air circulation in a building driven by the combination of
buoyancy of heated and fall-off by cooled air is a new application. The dimensioning of
all air leading parts in the greenhouse or building, especially height and diameter of the
tower, and the design of the heating and cooling system, mainly consisting of the heat
exchanger and the heat storage, are non trivial due to the complicated network of thermal
interactions, that characterizes the system. Some examples:
– The efficiency of the heat exchanger has an important influence on the temperature
difference of the air between bottom and top, thus on the air flow in the greenhouse.
– Increasing heat transfer in the heat exchanger enlarges the contact surface for the air
that causes higher pressure losses and a reduction of air mass flow.
– High relative humidity reduces heat losses by lowering the air temperature but can
decrease the air flow due to lower temperature differences between top and bottom.
– Solar energy can be used to increase the sensible or the latent heat of the air (or both).
High temperature and low relative humidity result in high thermal losses and
increased air mass flow with lower water content in the air.
– Increasing greenhouse temperatures result in increasing storage temperatures which
permit a smaller storage volume.
In addition the optimum air conditions for fruit production must be considered and
have a strong impact on the design of the heating and cooling system. Furthermore the
economical optimum between the value of the harvest over a year and of the value of the
saved water as well as the necessary investment costs must be accomplished.
There were many different investigations in the development of simulation tools.
Most of them are based on a mathematical representation of real systems using physical
rules. Models can become very large, particularly when local discretisation is used,
depending on the complexity of the represented system. Another program type is based on
experimental data. They determine the future system behaviour applying knowledge of
past behaviour. The advantage of these data based models is the mostly less complicated
structure, but in return they offer no insight into the system. The influence of
modifications in the construction of a real system or in the included technical devices can
not be determined. The intent to develop a planning tool is the reason for the
concentration on physically based models in this study. Programs that can be used in this
context have to accomplish five basic specifications. These are:
– the calculation of the air flow in the greenhouse,
– the thermal analysis of the construction and the enclosed air,
– the water balance including evapo-transpiration, air humidification, and condensation,
– the possibility of simulating the thermal dynamic of the technical equipment,
– and the ability to be adapted to unusual greenhouse geometries.
The investigations concerning existing simulation programs lead to a classification of the existing or available programs (Fig. 3). Most of them do not cover all five
abilities described, but include very helpful details in modelling specific aspects.
Computational Fluid Dynamics
Computational fluid dynamics (CFD) has been applied in various fields. It is used
to predict air flows in room, towers, greenhouses etc. (e.g. Lee et al., 2003). Primarily
used programs are FLUENT, Star-CD and CFX. The principle of CFD is the solution of a
set of non-linear partial differential equations which represents the fundamental physical
laws of the conservation of mass, momentum, and energy. Additionally the consideration
of turbulences that normally exist in most practical cases complicates matters
considerably. A common approach is the use of the Reynolds averaged form of the
Navier-Stokes equations, where the instantaneous air flow is assumed to be the sum of a
mean and a fluctuating component.
Thermal Analysis
Thermal analysis programs often originate from the simulation of buildings or
systems for heating, ventilation, and air conditioning (HVAC). A wide range of building
simulation programs is available, the most known are BLAST, DOE-2, TRNSYS, ESP or
TAS. A very informative site can be found at Most of them use
non-geometrical models and concentrated parameters for the included issues like walls,
windows, air etc. Programs for the simulation of HVAC systems including solar heating
and cooling are very often developed for the simulation of specific systems. But there are
also modular and flexible HVAC programs which provide a library of reusable components, the most known are TRSNYS and Smile (Nytsch-Geusen, 2001). Both programs
include an extensive library with models for buildings and HVAC systems. Their common
origin is the simulation of thermal solar systems that has evolved to the current main
application, the simulation of buildings.
Greenhouse Simulation and Plant Modelling
The simulation of greenhouses including plants can be done by means of steady
state models. The dynamical thermal effects like they are known in buildings or heat
storages are normally negligible here. There are many approaches to simulate
greenhouses, e.g. HORTITRANS (Jolliet, 1994) and HORTICERN (Jolliet et al., 1991).
An impressive overview on modelling solar energy input in greenhouses is given in
(Pieters and Deltour, 1999). Aspects of the convective heat transfer in greenhouses are
described in (Roy et al., 2002).
Pressure Node Models
Besides the more known method for the solution of the partial differential
equation system for the 2- or 3-dimensional flow used in CFD, the air flow can also be
determined by a much easier but less adequate method, the pressure node model. In this
kind of model the flow distribution inside a zone is not analysed, only the outgoing and
incoming flows at the borders of the homogeneous zone are calculated. This reduces the
equation system to a one-dimensional system. The state variables of the air in every zone,
like temperature, humidity, pressure, and pollutant concentration, are calculated by a
model using lumped parameters. Through defined air paths, e.g. windows, doors, cracks
in walls or windows, air change can occur. The air movement between the zones is
calculated by pressure balances for every zone considering the pressure drop passing the
air path resistance. Using air mass conservation for the complete network of pressure
nodes, a system of non linear equations is built and solved. A not very new but excellent
survey of airflow models in multi-zone structures is given in (Feustel and Dieris, 1992).
More recent analyses are described in (Dorer et al., 2001) and in (Nytsch-Geusen, 2001).
Aspects of Coupling CFD and Thermal Analysis
The use of CFD requires the proper definition of the boundary conditions of the
examined volume, including all interior wall temperatures, heat flows through the outer
areas and the incoming or outgoing air flows. These boundary conditions are extrapolated
into the cell grid in the investigated volume to solve the transport equations. Erroneous
boundary conditions lead to incorrect flow predictions. Considering the changing thermal
situations the boundary conditions have to be recalculated and the CFD analysis has to be
redone for the new situation. This produces long term analyses and is, until now, not
suitable as a design tool in the sense it is needed here.
Nevertheless there are some investigations (Negrao, 1995; Nytsch-Geusen et al.,
2003) to connect thermal analysis programs with CFD programs. Basically two different
coupling mechanisms are used. They differ in the coupling mechanism in one direction
(from thermal analysis to CFD) and in two directions (boundary conditions to CFD and
flow conditions back to thermal analysis). Depending on the time step in the thermal
analysis or the fact of appreciable changes in the inner surface conditions this can be done
related to special handshake schedule (Beausoleil-Morrison, 2000). The known investigations achieved interesting results, but related to model limitations, inconvenient use of
programs and extensive calculation times they could not yet establish new applicable
The simulation environment Smile (Ernst et al., 1997; Ernst et al., 2000) has been
chosen for the Watergy project. Smile has been developed by the Technical University
Berlin and the Fraunhofer Institute for Computer Architecture and Software Technology
since 1990. It is an universal simulation environment for time continuous and discrete
system modeling and is characterized by the following fundamental properties: equation
and object oriented modelling language, open and expandable, large model library, and
interchangeable numerical solvers with variable time steps.
The model specifications were written in the Smile modelling language. The
Smile model compiler translates this code into Objective-C and another C-compiler into
an executable program code. The model library contains a set of basic model classes for
building elements (walls, windows, roof, etc.), and for HVAC technical equipment (solar
collector, heat storage, heat exchanger, pumps, etc.). These models allow the configuretion of complex building models using three object oriented paradigmas encapsulation,
inheritance and aggregation. Besides the capacity of thermal building simulation the
building model can determine the interzonal air and moisture change through pressure
networks. Every zone can be equipped with a thermal, radiative and a pressure node.
Combining all zones to the simulation model allows to solve the resulting equation
system. Before Smile was applicable for the simulation of the Watergy greenhouse (prototype I) some model adjustments had to be done.
Exemplary Model Adaptions
1. Short Wave Radiation. The building elements in the model library were originally
developed for the modelling of „normal“ buildings with normally sized windows. For this
reason the window model code contains a complete transformation of direct radiation into
diffuse radiation. The diffuse radiation hits the inner surfaces (walls, roof, etc.) according
to their areas and absorption coefficients. This produces a negligible error in „normal“
buildings especially if the thermal inertia is high. In the case of greenhouses and
particularly in the Watergy situation where even two transparent layers have to be passed
sequentially, this assumption is no longer true. Considering in reality the existence of
direct short wave radiation inside the zone, the model had to be changed by the
introduction of a weighting factor which allocates a main direction to the „diffuse“
2. Long Wave Radiation. The models for windows normally assume glass as transparent
material. Unlike plastic foils glass is a (almost) perfect barrier for long wave radiation. As
it is very common to use plastic foils for the transparent cover of greenhouses the window
model had to be changed to include the transmission of long wave radiation.
Vegetation. For the plant model a new Smile class had to be defined. The
mathematical base has been adapted from the FAO Irrigation and Drainage Paper, No. 56.
Using the inheritance feature of Smile additional crop models can easily be added.
Air humidifier. As mentioned before the moving air is going to be humidified
above the inner roof using sprinklers. The required model is very sensitive for the
unknown droplet size in the sprays. This is why the model will have to be based on
empirical data after the measurements.
For the simulation of the prototype 6 zones were chosen (Fig. 4). The bottom zone
with the plant area, the top zone between the two roofs, the tower zone outside the
chimney and three further zones inside the chimney including the heat exchanger. Fig. 5
shows the used surfaces for the definition of the bottom zone as an exemplary
development view. The areas in use are defined by their physical properties and their
orientation. of the greenhouse.
The forgoing provides an insight into the difficulties of simulating closed
greenhouses with natural convection and facilities for heating and cooling. There are
various approaches for parts of the in this case examined systems, speaking of
greenhouses, plants, air movements or the technical equipment like heat exchanger and
However for the prediction of the behaviour of a combined apparatus in the sense
it is described here it is essential to implement the complete system because of the strong
interactions between the different elements. Closed systems with natural ventilation
suggest a system analysis with CFD methods. But in the context of the development of a
design and planning tool long term simulations must be considered that cannot be done
with CFD. One approach to combine fluid dynamic analyses with thermal simulations
consists of using of so called pressure node models. These models simplify the air flow to
a 1-dimensional problem at the connections between two different zones of the model.
The ability to determine the air flow with the pressure node method, the thermal
behaviour of the building including plants, and the HVAC system could be found in the
simulation environment Smile. The existing model library could be adapted and
extended. The first evaluation of the measurements are expected April/May 2005. At that
point a comparison of calculated and real data can begin.
This research is funded by EESD, the EC 5th Framework Programme promoting
Energy, Environment and Sustainable Development (Project Number NNE5-2001-683).
Literature Cited
Beausoleil-Morrison, I. 2000. The Adaptive Coupling of Heat and Air Flow Modelling
within Dynamic Whole-Building Simulation. PhD Thesis. University of Strathclyde.
Glasgow, UK.
Buchholz, M., Jochum, P. and Zaragoza, G. 2005. Concept for Water, Heat and Food
Supply from a Closed Greenhouse - The Watergy Project. Acta Hort. 691:509-516.
Dorer, V., Haas, A., Keilholz, W., Pelletret, R. and Weber, A. 2001. COMIS simulation
environment for multizone air flow and pollutant transport modelling. Seventh
International IBPSA Conference. Rio de Janeiro, Brazil 13-15 August. p. 403-410.
Ernst, T., Jähnichen, S. and Klose, M. 1997. The architecture of the SMILE/M simulation
environment.15th World Conference on Science Computation, Modelling and Applied
Mathematics. Berlin.
Ernst, T., Klein-Robbenhaar, C., Nordwig, A. and Schrag, T. 2000. Modeling and simulation of hybrid systems with SMILE. Informatik, Forschung und Entwicklung.15: 33–
Feustel, H.E. and Dieris, J. 1992. A survey of airflow models for multizone Structures.
Energy and Buildings. 18: 79-100.
Janssen, H.J.J., Speetjens, B., Stigter, H., van Straten, G. and Gieling, Th.H. 2005.
Watergy, Infrastructure for Process Control in Semi-arid Regions. Acta Hort. 691:821828.
Jolliet, O. 1994. HORTITRANS, A Model for Predicting and Optimizing Humidity and
Transpiration in Greenhouses. J. Agric. Eng. Res. 57: 23-37.
Jolliet, O., Danloy, L., Gay, J.-B., Munday, G.L. and Reist, A. 1991. HORTICERN. An
improved static model for predicting the energy consumption of a greenhouse.
Agricultural and Forest Meteorology. 55:265-294
Lee, I., Kang, C., Yun, J., Jeun, J. and Kim, G. 2003. A Study of Aerodynamics in Agriculture. Forum on Bioproduction in East Asia: Technology Development and
Opportunities. ASAE Annual Meeting. LasVegas, USA. 27 July.
Negrao, C.O.R. 1995. Conflation of Computational Fluid Dynamics and Building
Thermal Simulation. PhD Thesis. University of Strathclyde. Glasgow, UK.
Nytsch-Geusen, C., Klempin, C., Nunez v. Voigt, J. and Rädler, J. 2003. Integration of
CAAD, thermal building simulation and CFD by using the IFC data exchange format.
8th International IBPSA Conference. Eindhoven, NL, p. 967-973
Nytsch-Geusen, C. 2001. Calculation and optimization of the energy efficiency of
buildings and their energy within an object oriented simulation environment. PhD
thesis, University of Technology Berlin, Germany.
Pieters, J.G. and Deltour, J.M. 1999. Modelling solar energy input in greenhouses. Solar
Energy. 67(2).
Roy, J.C., Boulard, T., Kittas, C. and Wang, S. 2002. Convective and Ventilation Transfers
in Greenhouses, Biosystems Engineering. 83(1): 1-20.
Fig. 1. Energy flow in the humid air collector (1: greenhouse, 2: tower for buoyancy,
3: cooling duct, 4: diurnal heat storage, 5:building, 6: seasonal heat storage).
Fig. 2. Water cycle in the humid air collector (1: collected water , 2: irrigation water,
3: water buffer, 4: additional humidification, 5: water use in the building).
Fig. 3. Capacities of existing simulation programs to predict the overall behaviour of
closed greenhouses or parts of it.
Division of the Greenhouse in Zones
zone 3
zone 4, 5 and 6
inside the
returning duct
with the heat
zone 2
zone 1
Fig. 4. The greenhouse is divided in 6 zones. Every zone has its own nodes for temperature, radiation and pressure.
door (air path)
wall 2
tower (air path to zone
window 3
window 2
window 1
inner roof
wall 1
window 4
wall 3
zone 2 (air path)
(complete roof)
wall 4
Fig. 5. Building elements of the plant zone as a developed view. For every element the
material and surface temperatures were calculated and their influence on the
enclosed air determined. Due to the non geometric model the location of an
element has no influence on the inner air temperature. “Walls” represent the steel
construction of the greenhouse.