Fuzzy Control of a Water Pump for an Agricultural Plant Growth System

Jos´e Dias

, Jo˜ao Paulo Coelho

1,2

and Jos´e Gonc¸alves

1,2

Instituto Polit´ecnico de Braganc¸a, Escola Superior de Tecnologia e Gest˜ao, Campus de Sta. Apol´onia,

5300-253 Braganc¸a, Portugal

INESC TEC Technology and Science, Campus da FEUP, 4200 - 465, Porto, Portugal

Keywords:

Renewable Energy, Fuzzy Controller, Agriculture.

Abstract:

At the present time there is a high pressure toward the improvement of all the production processes. Those

improvements can be sensed in several directions in particular those that involve energy efﬁciency. The deﬁ-

nition of tight energy efﬁciency improvement policies is transversal to several operational areas ranging from

industry to public services. As can be expected, agricultural processes are not immune to this tendency. This

statement takes more severe contours when dealing with indoor productions where it is required to artiﬁcially

control the climate inside the building or a partial growing zone. Regarding the latter, this paper presents an

innovative system that improves energy efﬁciency of a trees growing platform. This new system requires the

control of both a water pump and a gas heating system based on information provided by an array of sen-

sors. In order to do this, a multi-input, multi-output regulator was implemented by means of a Fuzzy logic

control strategy. Presented results show that it is possible to simultaneously keep track of the desired growing

temperature set-point while maintaining actuators stress within an acceptable range.

1 INTRODUCTION

Nowadays, in order to increase the companies com-

petitiveness, all the production processes are subject

to rigorous audits. The output of such studies lead to

changes in several key variables that will, ultimately,

lead to a performance increase. Those changes can

happen at several levels, ranging from plant layout,

tasks scheduling, rework or scrap minimization and,

of course, energy efﬁciency.

Energy efﬁciency is a pressing problem that is ad-

dressed at severallevels. Inparticular,at the European

Union, there are currently several political mecha-

nisms undergoing in order to both reduce the energy

consumption and also to promote renewable-based

energy production systems (Armstrong and Blundell,

2007).

However economic growth is tightly connected to

energy consumption. Hence, in order to cope with all

those constraints one must be able to ﬁnd alternative

strategies to devise more energy efﬁcient production

methods that will not scale up with the required eco-

nomic development.

As can be expected, agricultural processes are not

immune to this tendency. Within a global market

paradigm, the costs reduction while maintaining the

products quality, is a constant grower concern. Hence

maintaining market competitiveness can then be a

challenge tackled at both operational and technologi-

cal fronts.

Regarding the latter, often it is possible to trans-

pose solutions used in different contexts to alternative

application bringing value to a new process. For ex-

ample the water heating systems, based on thermody-

namic pumps used in domestic applications, can be

also used in agricultural processes as is the case of

greenhouses productions.

This paper deals with a new devised strategy to

improve the energy efﬁciency of a tree nursing sta-

tion. This new solution replaces the actual resistor

based heating system by an alternative using circula-

tion of hot water in a closed pipe circuit. The water

recirculation is performed by an electric pump and the

heating is taken from the environment by means of a

set of thermodynamic panels.

However, prior to the effective implementation of

this solution, it is required to test if this alternative

heating system can, in fact, be used to replace the old

one. In order to do this a scaled version of a growing

platform was built where the thermodynamic heating

system was replaced by a butane gas water heater.

Both water pump and gas heating system must be

156

Dias, J., Coelho, J. and Gonçalves, J..

Fuzzy Control of a Water Pump for an Agricultural Plant Growth System.

In Proceedings of the 7th International Joint Conference on Computational Intelligence (IJCCI 2015) - Volume 2: FCTA, pages 156-161

ISBN: 978-989-758-157-1

controlled taking into consideration the imposed tem-

perature set-point while maintaining system integrity.

The actuators state will be function of exogenous in-

formation provided by an array of sensors and the

control law. In this work a Fuzzy controller strategy

was pursued due mainly to the fact that, under this

paradigm, a expert type controller is easily translated

to a Fuzzy behaviour system (Kia et al., 2009).

The remain of this paper will be divided into four

additional sections. After a more thorough presenta-

tion of the addressed problem at section 2 the Fuzzy

controller design strategy will be described along sec-

tion 3. The obtained results, regarding both set-point

tracking and actuator wearing, will be presented at

section 4. Finally some concluding remarks, as long

as future work trends, will be presented in section 5.

2 PROBLEM DESCRIPTION

In this section the addressed problem will be dis-

cussed in further deep. In order to do that some previ-

ous contextualization about the actual installed trees

growing system will be provided. This system con-

sists on platforms with around six meters of length

and near two meters width placed inside a greenhouse

where the air temperature is roughly controlled. In

particular, the inside air temperature ranges, in aver-

age, from 10 to 30 degrees centigrade.

At the present time the installed growing machine

is used to nurse olive trees and chestnuts, among oth-

ers, until they are sufﬁciently resistant to be planted

outdoors (Jacobs et al., 2009). The system is a tray,

covered with perlite, and with an electric resistance

coil format, powering 3 kW, running along the cov-

ered area. Actually the greenhouse has 9 growing

stands leading to an installed power of near 30 kW.

A picture of the above referred growing platform sys-

tem is depicted in Figure 1.

During the winter time, when the indoor temper-

atures are lower, the average energy spent in each

growing platforms is, in average, around 60 kWh per

day. This number is excessive and import high pro-

duction costs. In order to reduce this value an alterna-

tive heating system was devised by replacing the in-

efﬁcient resistance based heating element by an water

heating system drive by thermodynamic panels (Yang

et al., 2007) (Huiling and Xiangzhao, 2012).

The heating process is straightforward as can be

exempliﬁed in the diagram of Figure 2.

Before implementing this heating system, and

since the heat and ﬂow requirements of the growing

stand, under nominal and extreme conditions, is un-

known, a scale prototype was built. First the stand

Figure 1: The currently installed growing platform inside

the greenhouse. The center pipes are for irrigation purposes.

Figure 2: Block diagram of the growing stand heating sys-

tem.

was implemented using the same structure material

as the original. Only its dimensions was scaled to a

platform of 125 cm × 65 cm. The height of 18 cm

was deliberately let equal to the original in order for

the solution devised for the prototype to be easily in-

tegrated thereafter in the real system.

The piping system was embedded in a special

modelled styrofoam as shown in Figure 3. Then this

assembly was placed inside the metal frame and ﬁlled

with perlite.

In the test rig developed, the thermodynamic heat-

ing system was replaced by a butane gas heating. This

operating change was mainly due to controllability

and logistics. Nevertheless, the information one seeks

to ﬁnd could be also easily obtained by this new ap-

proach. Figure 4 shows an image of this alternative

structure.

Moreover an array of sensors was dispersed along

the system in several key points. Six waterproof tem-

perature sensors (DS18B20) were distributed along

the perlite surface. This set of sensors allows to anal-

yse how well and uniform the heat spreads along the

working area. In addition other temperature sensors

are available: one to measure the heater outlet water

temperature, the water temperature at a reservoir and

the environment air temperature. The water ﬂow is

also measured by means of a ﬂow sensor.

The water is made circulating by means of a 1/2

HP water pump system as can be observed from Fig-

Fuzzy Control of a Water Pump for an Agricultural Plant Growth System

157

Figure 3: The growing stand internal piping system mean-

der around a piece of styrofoam used in radiant ﬂoors do-

mestic heating systems.

Figure 4: Distribution of the temperature sensors along the

perlite.

ure 5. The water pump is connected to the gas heater

system by means of 1 inch pipes. On the other hand,

the heater outlet pipe feeds the growing table. The

circuit is closed by means of a return pipe from the

table to the reservoir.

The control and data acquisition system was built

around an Arduino Uno R3.0 (Dhamakale and Patil,

2011). Most temperature sensors connects to the mi-

crocontroller board by means of One-Wire communi-

cation protocol. Others require the use of an analog

to digital (A/D) input. The time is kept tracked us-

ing a real-time clock (RTC) and data is locally saved

in a secure digital (SD) memory card. The former

is connected to the Arduino platform by means of

inter-integrated circuit communication protocol (I

and the latter by serial peripheral interface (SPI). This

hardware setup can be observed from Figure 6.

The trees nursing process requires that the surface

Figure 5: Detail on the water circulating system. Below the

table it is possible to observe both the water pump and the

reservoir.

Figure 6: Control and data acquisition hardware built

around an Arduino Uno board.

temperature be maintained around 23 degrees centi-

grade. In this process the main external disturbances

are the indoor air temperature ﬂuctuations and the

load imposed by periodic irrigations. However, in this

work, the latter is not considered.

In order to cope with the requiredsurface tempera-

ture set-point two manipulated variables are available.

The heater and the water pump states. Both have an

important role in temperature regulation that will be

further explained in the next section along with the

control law devised to regulate this system.

3 THE FUZZY CONTROLLER

Besides set-point tracking, the addressed system also

requires water temperature supervision in order to

FCTA 2015 - 7th International Conference on Fuzzy Computation Theory and Applications

158

prevent the system collapse. Moreover, due to the ac-

tuators nature, a bang-bang approach to control must

be performed. The use of an approximated ﬁrst or-

der system with low bandwidth, a classical approach

to controller system design should not be a challenge.

However, and in order to approximate de controller

behaviour by an expert controller point-of-view, a

Fuzzy based controller was devised (Dhamakale and

Patil, 2011).

The controller design was carried out using the

MATLAB



Fuzzy Logic Toolbox taking into consider-

ation empirical knowledge. The controller structure,

within the Fuzzy Toolbox context, can be observed

from Figure 7.

This is a multi-input, multi-output controller sys-

tem whose inputs are the average perlite surface tem-

perature and the water temperature inside the reser-

voir. The output variables are the heater and water

pump states.

The fuzzy inference mechanism selected was a

Mamdani type (Mamdani and Assilian, 1975; Sakti,

2014). Moreover the conjunction and disjunction

rules operation were the minimum and maximum re-

spectively. The defuzziﬁcation process is carried out

by means of a centroid operation over the feature

space.

Figure 7: Overall Fuzzy controller structure and

parametrization.

Regarding the surface temperature variable, the

input space was partitioned into ﬁve Gaussian type

Fuzzy sets. Each one was labelled as ‘cold’, ‘cool’,

‘good’, ‘warm’ and ‘hot’. This fragmentation can be

depicted from Figure 8.

Regarding the reservoir water temperature, the in-

put space was divided into three sets labelled ‘cold’,

“good’ and ‘hot’. The sets distribution along the input

space range can be observed from Figure 9.

Now the output space for the water pump was split

into three triangle type membership functions as can

be concluded from Figure 10.

The same operation was carried out for the heater

Figure 8: Surface temperature fuzziﬁcation. This input vari-

able was differentiated into ﬁve partially overlapped sets.

Figure 9: Reservoir water temperature fuzziﬁcation. Three

sets was used to describe the water temperature status over

the range from 10 to 65 degrees centigrade.

Figure 10: Output space partition for the water pump using

three triangle type membership function.

variable. However, in this case, a ﬁve membership

functions set was used to describe the output space.

Nevertheless the same triangular shape membership

functions were used as in the previous case as can be

seen from Figure 11.

The Fuzzy controller rules were produced auto-

matically by the MATLAB



software. In particular

a total of 15 rules were produced. This rule based

controller structure can be easily analysed by means

Fuzzy Control of a Water Pump for an Agricultural Plant Growth System

159

Figure 11: Output space partition for the gas heating system

using ﬁve triangular shape membership functions.

of a graphical output available from the toolbox. A

screenshot of this rule viewer graphical tool is pre-

sented in Figure 12.

Figure 12: The rule viewer assuming an surface temperature

of 23 degrees centigrade and a reservoir water temperature

of 37.5 degrees centigrade.

In the next section the above designed Fuzzy con-

troller behaviour will be presented. Its performance

will be analysed regardingboth set-point tracking and

actuators stress. Even if the former ﬁgure of merit can

be slightly relaxed the latter is of extreme importance

in order to reduce wearing of both the water pump and

the gas heating system by preventing high frequency

switching states.

4 RESULTS AND DISCUSSION

Several tests were performed, always giving priority

to the worst case scenario, with ambient temperatures

close to 10 degrees centigrade. In all the performed

tests the desired set-point temperature is considered

constant and equal to 23 degrees centigrade. Below

are presented the results of two tests. The ﬁrst, whose

results are illustrated in Figure 13, has a temperature

starting point of around about 16 degrees centigrade.

Figure 13: Results obtained during the ﬁrst test. From top

to bottom one can observe the following curves: Deposit

temperature, set-point, average temperature, ambient tem-

perature and heater status.

A second experiment, with lower water tempera-

ture starting point, was also performed. In this second

case the initial system temperature is approximately

14 degrees centigradeand the averagetemperature en-

vironment stays around 11 degrees centigrade. The

obtained results are presented on Figure 14.

Figure 14: Results obtained during the second test. From

top to bottom one can observe the following curves: De-

posit temperature, set-point, average temperature, ambient

temperature and heater status.

From both experiments it is possible to conclude

that, in both cases, the controller was able to reach,

and maintain, the set-point temperature. Moreover

one can see that the closed loop system exhibits an

over-damped response type. In addition it is possible

to observe the heater state change during the opera-

tion. Notice that the water pump state did not change

during the experiments and was always on. From the

obtained data it is possible to conclude that the rate of

change of the heater state has a period of around 30

minutes. Hence the designed controllerdid not lead to

an aggressive closed loop response from the actuators

point-of-view.

FCTA 2015 - 7th International Conference on Fuzzy Computation Theory and Applications

160

5 CONCLUSION

This paper addresses the implementation of a Fuzzy

controller for a new tree nursing station heating sys-

tem based on thermodynamic panels. This alternative

strategy, if proven economically viable, will replace

the older system whose heating system is based on

electric heating elements. In order to infer the appli-

cability of this technique a scaled test rig was built.

This test rig was used to test if it was possible to at-

tain the set-point temperature in the worst operating

conditions and, if so, what will be the water tempera-

ture required and its mass ﬂow. In the test rig the wa-

ter was heated, not using the thermodynamic heating

system, but a gas boiler whose state could be easily

controlled and minimizing logistic problems. In addi-

tion, a half horse power water pump was installed, in

order to make the water circulating. The pump opera-

tion was controlled by a simple on-off strategy.

In order to maintain the system state variables in

the required range a multi-input, multi-output Fuzzy

controller was designed. The controller input are the

temperature information collected by an array of sen-

sors and its output signal are the commands for both

the water pump and the gas heater. The choice of this

controller paradigm was due mainly to its simplicity

and its proximity to empirical control based on expert

knowledge.

From the obtained results it is possible to con-

clude that the designed controller was able to main-

tain the average surface temperature at the set-point

level without generating any high frequency control

signals. This is very important in order to reduce

the wearing inherent to rapid on-off actuator state

changes.

However further tests must be still performed us-

ing different operating conditions like increased air

temperature, adding moisture to the perlite substrate

since, in real operating conditions, the irrigation sys-

tem periodically sprinkle the small trees. It is neces-

sary to analyse how this change in thermal conductiv-

ity will be noticed in the overall system performance.

ACKNOWLEDGEMENTS

This work is ﬁnanced by the ERDF - European Re-

gional Development Fund through the COMPETE

Programme (operational programme for competitive-

ness) and by National Funds through the FCT -

Fundac¸˜ao para a Ciˆencia e a Tecnologia (Portuguese

Foundation for Science and Technology) within

project “*FCOMP-01-0124-FEDER-037281*”.

REFERENCES

Armstrong, F. and Blundell, K. (2007). Energy - beyond oil.

Oxford University Press.

Dhamakale, D. and Patil, S. (2011). Fuzzy logic approach

with microcontroller for climate controlling in green

house. International Journal on Emerging Technolo-

gies, 2:17–19.

Huiling, Z. and Xiangzhao, F. (2012). An experimental re-

search on the application of ground source heat pump

and ﬂoor-heating system in the greenhouse. Power

and Energy Engineering Conference, pages 1–4.

Jacobs, D. F., Landis, T. D., and Luna, T. (2009). Nursery

manual for native plants. R. Dumroese et al.

Kia, P. J., Far, A. T., Omid, M., Alimardani, R., and Nader-

loo, L. (2009). Intelligent control based fuzzy logic for

automation of greenhouse irrigation system and eval-

uation in relation to conventional systems. World Ap-

plied Sciences Journal, 6(1):16–23.

Mamdani, E. and Assilian, S. (1975). An experiment in lin-

guistic synthesis with a fuzzy controller. nternational

Journal of Man-Machine studies, 7:1–13.

Sakti, I. (2014). Methodology of fuzzy logic with mamdani

fuzzy models applied to the microcontroller. Informa-

tion Technology, Computer and Electrical Engineer-

ing.

Yang, Z., Pedersen, G., Larsen, L., and Thybo, H. (2007).

Modeling and control of indoor climate using a heat

pump based ﬂoor heating system. IECON 2007 - 33rd

Annual Conference of the IEEE Industrial Electronics

Society, page 29852990.

Fuzzy Control of a Water Pump for an Agricultural Plant Growth System

161