Professional Documents
Culture Documents
Description of a lighting scenarios in the sense that it ers are used is one answer to the grow-
increases the length of the period of
Crop Growth supplemental lighting when the daily
ing information-processing require-
ments (Bakker et al., 1988; Hooper,
solar light integral is low, and reduces
Model for the or eliminates the use of supplemental
1988).
lighting when the weather forecast A working group in the Nether-
Management of predicts that the daily solar light inte- lands recognized the important role
gral will exceed plant requirements. that can be played by optimization
Supplemental calculations within dynamic climate
management strategies. They proposed
Lighting in an approach based on several levels of
I
n greenhouse production, the cre-
Greenhouses ation of an artificial climate often
control (Udink ten Cate et al., 1978).
Decisions are based on output from
requires large energy inputs. Care- specialized simulation models (Challa
ful management of energy costs re- and Schapendonk, 1986; Heuvelink
Michel Carrier1, quires the use of computerized sys- and Challa, 1989), and may be subject
André Gosselin1, and tems that continuously monitor and to considerations involving economic
change important climatic parameters. optimization and user preferences
Laurent Gauthier2 The operation of the current genera- (Challa et al., 1988). The environ-
tion of commercial climate control mental conditions are then optimized
systems is based on the observance of over short time intervals (a few min-
Additional index words. Lycopersicon production guidelines (or blueprints) utes).
esculentum, object-oriented program- set by the producer for relatively long At Laval Univ., a dynamic green-
ming, expert systems, climate control, periods (several days). These guide- house climate management software
simulation lines are generally the result of the system has been developed. The sys-
experience of hundreds of producers tem, called GX (Gauthier, 1993a,
Summary. A dynamic management over several years. The guidelines aim 1993b), defines a general architecture
strategy for supplemental lighting in at maintaining climatic conditions that
greenhouses was developed. It makes
that may accommodate different deci-
ensure a good growth rate and avoid sion-support or decisionmaking mod-
use of a plant growth model and of a extreme conditions.
rule-based decisionmaking protocol ules such as mathematical models and
within the framework of a generic
Although very useful, existing sys- rule bases. GX can be used as an on-line
greenhouse climate management soft- tems do not maintain optimal growth system (i.e., for the real-time manage-
ware system. The model, an adapted conditions at all times, and the particu- ment of greenhouse climates). It also
version of SUCROS87, tracks plant lar needs of an individual producer are can be used in a totally virtual mode to
growth and predicts dry weight pro- not always met by the functions pro- simulate and assess the effect of various
duction based on measured or esti- vided by the control system. Further, management strategies. To that end, it
mated values of light intensity, temp- these systems are essentially static contains models of the mass and energy
erature, and CO2 concentration. A set (Udink ten Cate, 1980) and cannot
of logical predicates (rules) imple- exchange processes that are used to
take into account the continuous simulate the greenhouse climatic envi-
ments the strategy’s behavior. Opti- changes affecting the greenhouse en-
mization of lamp use was conducted ronment and its response to the control
as a function of economic criteria that
vironment. Challa and Schapendonk strategy under study.
enable a comparison between the ad- (1986) showed that the optimal cli- This article presents results ob-
ditional income associated with yield matic set points vary over short periods tained with GX following the develop-
increases due to supplemental lighting of time (within the same day) as a ment and simulated use of a supple-
and incurred cost increases. Although function of both outside climatic pa- mental lighting management strategy.
the model is not perfectly reliable in rameters and plant development stage. This strategy is based on the use of a
its predictions, the system can be used Further, the expenses associated with general model of plant growth,
to simulate the effect of changes to maintaining the settings varies with
economic parameters on the decisions SUCROS87, that was adapted to the
season and cost of inputs. use of artificial lighting, and on a series
of the management strategy. The dy- Research on more-flexible con-
namic strategy described here differs of management rules supplied by a
trol systems has led to the study and domain expert (i.e., a grower that has
development of dynamic systems based several years of experience working
1
Département de phytologie, Centre de recherche en hor- on explanatory models of growth and with supplemental lighting). The num-
ticulture, Faculté des Sciences de l’Agriculture et de development and on models of green-
l’Alimentation. Université Laval, Sainte-Foy, Québec, ber of rules was kept at a minimum in
Canada, G1K 7P4. house mass and energy balances (Challa order to limit the overall complexity of
2
Département de génie rural, Centre de recherche en et al., 1988 ; Gauthier and Guay, 1990). the knowledge base and to facilitate
horticulture, Faculté des Sciences de l’Agriculture et de Researchers have recognized the modifications.
l’Alimentation. Université Laval, Sainte-Foy, Québec, need for multilevel control systems
Canada, G1K 7P4. This article describes the opera-
that can synchronize the reactions of tion and use of a crop growth model
We thank Le Fonds FCAR, the National Science and
Engineering Research Council, and Hydro-Quebec for
system components to the producer’s and of management heuristics (infor-
their financial support. The cost of publishing this paper ultimate objective (Udink ten Cate et mal or learned behavioral knowledge)
was defrayed in part by the payment of page charges. al., 1978). These systems demand con-
Under postal regulations, this paper therefore must be for the operation of supplemental light-
hereby marked advertisement solely to indicate this fact. siderable processing power. A distrib- ing systems in greenhouses.
Table 1. Names and descriptions of variables used in the scripts and scenario for supplemental light management.
lamps until the end of the working day. night). The first part of the rule calcu- Precision of the
Rule 5 is thus triggered when the sun lates the latest time in the morning at
provides insufficient light and the cur- which the lighting can be turned on
SucrosTomato model
rent time is smaller than the and still respect minimal photoperiod To be useful for management pur-
EndOfWorkShift. The grower thus has duration, assuming that the lights will poses, a crop growth model must be
the potential of overriding the eco- be turned off at sunset or at the end of reliable. The weekly mean yield pre-
nomic optimization by setting the value the working day at the latest. dicted by the model was compared to
of the EndOfWorkShift parameter. This actual yields for a greenhouse tomato
rule originally was considered impor- RULE 6: crop grown between 31 Oct. 1990
tant. However, recent discussions with IF: CurrentTime >= (max(Sunset, and 8 Feb. 1991. The mean difference
growers leads us to reassess the useful- EndofWorkshift) – MinPhotoperiod) was 0.41 kg•m–2 per week, which is
ness of this rule, because both the AND: CurrentTime < Sunrise roughly equivalent to the difference
work patterns and lighting regimes THEN: DesiredPAR := 90 between yield under supplemental
have been modified to accommodate JUSTIFICATION: Lamps turned on lighting (at an intensity of 100
the needs of pollination bumble bees. to ensure a minimum photoperiod mmol•m–2•s –1 of HPS for a 17-h pho-
It appears that the use of supplemental toperiod) and yield without supple-
lighting past the time of sunset jeopar- The last two rules use the agent mental lighting for the same period of
dizes the bees’ ability to regain their SucrosTomato to determine the prof- the year (0.45 kg•m–2 per week) (Vézina
hives at the end of the day. itability of supplemental lighting. These et al., 1991). Hence, the precision of
rules have a lower priority than the the model is inadequate.
RULE 5: preceding rules and only serve when As GX makes decisions based on
IF: IntensitySolarRad < VisibleLight- none of the other rules are triggered. economic criteria, the profit associ-
Threshold ated with the use of supplemental light-
AND: CurrentTime < EndOfWork- RULE 7: ing should be greater than the model’s
Shift IF: PredictedProfit >= 0 error margin (converted into com-
AND: CurrentTime > Sunrise THEN: DesiredPAR := 90 mercial value). If not, a profit calcu-
THEN: DesiredPAR := 90 JUSTIFICATION: It is profitable to lated by SucrosTomato may be a loss
JUSTIFICATION: Keep lamps on use the HID lamps in reality.
until the end of the working day The profit (or loss) attributed to
RULE 8: the model’s error margin (Perr) is ex-
Rule 4 establishes the maximum IF: PredictedProfit <= 0 pressed in $/ha and calculated by sub-
photoperiod duration. Large variations THEN: DesiredPAR := 90 tracting the packaging costs associated
in photoperiods may not be beneficial JUSTIFICATION: It is not profitable with the quantity of fruit produced
for the plants. Thus, rule 6 ensures a to use the HID lamps from the gross revenue from product
minimal photoperiod duration based sales [Eq. 1]. The additional cost of
on the parameter MinPhotoperiod. In In GX, it is the scenario that con- electricity is not included because the
the province of Quebec, the electricity sults the rule base. The rule base is extra yield due to overestimates by the
rates can make it more economical to consulted at regular intervals (i.e., ev- model does not depend on the quan-
use supplemental light in the morning ery 15 min) and only the rules for tity of energy used.
(from midnight to 0600 HR) than dur- which the premises are true are trig-
ing the evening (1600 HR until mid- gered. Perr = E * H * (1/ρ) * (P – C pack) [1]
Fig. 3. Simulated effect of the static and dynamic control strategies on PAR light levels for 2 Jan. 1990. The dynamic strategy used supplemental
light for as long as it was judged profitable.
the behavior of the two strategies for ing predicted light levels and a cost– the night. In fact, because the lamps
the conditions measured on 2 Jan. benefit analysis approach to decision- contribute significantly to greenhouse
1990 is compared. The light energy making. The dynamic nature of the heating (Brault et al., 1989), their
for the day was low and the instanta- optimization strategy is also well-illus- effect on the greenhouse heat balance
neous maximum intensity (PAR in the trated in Fig. 4. As the amount of should be taken into account when
greenhouse) seldom exceeded 100 anticipated and received solar radia- decisions are being made about chang-
W•m–2. Under these conditions, the tion increases, the number of hours ing the state of the lamps. Another
calculations by SucrosTomato showed during which supplemental PAR is aspect that should be considered for
that it was profitable to use supple- used decreases. The same trend occurs optimization purposes is the future
mental lighting for a large part of the with the photoperiod. worth of a crop’s net growth rate,
day. The conventional system turned For example, on 1 Mar., the lamps because the market price of produce
off the lamps at a predetermined in- were not turned on in the early part of such as tomatoes tends to vary signifi-
stantaneous intensity. In both cases, the day because the anticipated light cantly throughout the year and today’s
the lamps were turned on again at the integral was greater than MaxDaily- fruit growth is next month’s sale. A
end of the day to allow greenhouse Light. However, the lamps were turned model to predict the evolution of mar-
workers to continue working. on later in the day (at 1200 HR) be- ket prices is thus an important piece of
The sensitivity of the dynamic cause it was estimated to be profitable the optimization puzzle.
strategy can be appreciated from the to do so. On the next day (2 Mar. Many other environmental pa-
fact that for the conditions of 2 Jan. 1990), very little light was anticipated rameters affect crop growth and en-
1990, when the price of tomatoes was and the lamps were switched on at ergy consumption. For example, CO2
set at $4.00 /kg, the strategy calcu- 0130 HR and kept on all day. In fact, on and relative humidity levels have an
lated a profit and the lamps were used that day, the light integral was not impact on crop growth and can be
for >12 h. However, when the price reached. controlled or affected by ventilation.
was $2.00 /kg for the same day, the The variability in supplemental Other considerations linked to pest
HID lamps were not used because the PAR and photoperiod was much lower management, pollination by bumble
program determined that it was not with the static strategy because the bees, and marketing can come into
profitable to do so. For the same day, lamps always were switched on at the play and influence the decisionmaking
the program calculated a loss when the same time of day and were switched off protocol. Hence, decisions based solely
intensity of supplemental lighting was when the radiation was above the on profit vs. loss considerations cannot
fixed at 50 mmol• m–2 •s–1 (10 W• m–2 ), threshold value or at the end of the always be used.
whereas it predicted a profit at 100 day.
mmol•m–2•s –1 (20 W•m–2). Conclusion
Table 2 contains a summary of Discussion The simulation capabilities of the
the results obtained from the simula- As can be seen from the values GX system provides a powerful tool for
tion of the behavior of the two strate- presented in Table 2, the dynamic the development, validation, and analy-
gies for meteorological conditions mea- strategy could result, during certain sis of dynamic strategies. The results
sured at Laval Univ. between 28 Feb. parts of the year, in significant electri- presented here showed that the inte-
and 27 Mar. 1990. As can be observed, cal energy savings while maintaining gration of a growth model and man-
the dynamic strategy resulted in a the same or even better light regimes. agement rules can be used effectively
greater amount of PAR energy while It is not clear if variations in photope- for the dynamic management of supple-
using less supplemental lighting. The riod can seriously affect crop growth, mental lighting in greenhouses. True
dynamic nature of the optimization but, because the amplitude of these optimization could be realized by tak-
strategy is illustrated by the fact that variations can be limited by the green- ing into account the effect of HID
standard deviation values for photope- house manager, this should not pose lamps on the greenhouse sensible and
riod and hours of supplemental light- any serious problem. latent heat balances. In fact, the ap-
ing are greater in the dynamic strategy The strategy presented herein only proach used for this study, which con-
than in the static strategy. In other manages supplemental lighting. A sisted in the integration of heuristics
words, while the static strategy used a natural extension to the management and analytical models, holds promise
fixed photoperiod resulting in unpro- of supplemental lighting would be the for the future because considerable
ductive use of supplemental lighting, optimization of thermal/shading efforts are being put into the develop-
the dynamic strategy made more ratio- screen control because such screens ment and improvement of greenhouse
nal use of the electrical energy needed have an affect on light levels when used heat exchange and crop growth mod-
to operate the HID lamps by consider- during the day and on heat loss during els.
In other words, computers and
Table 2. Comparison of radiation integrals, supplemental light use, and photoperiods obtained
from simulations of the behavior of the static and dynamic strategies for meteorological their associated software can automati-
conditions measured at Laval Univ. between 28 Feb. and 27 Mar. 1990. cally and dynamically manipulate cli-
matic and environmental set-points.
Total PPFD (mol•m –2) HID usage (h) Photoperiod (h) This can result in significant reduc-
Static Dynamic Static Dynamic Static Dynamic tions in costs and in increases in pro-
Average 13.8 16.6 13.1 6.5 15.0 13.1 ductivity. In fact, we have shown how
Sum 386.4 465.5 366.5 180.8 419.5 366.2 software can be used to make better
SD 2.8 4.6 1.1 4.4 0.3 2.0 (read, more-profitable) use of existing
technology such as supplemental light-
ing. There are many opportunities for Challa, H. and A.H.C.M. Schapendonk. Goldberg, A. and D. Robson. 1983.
such improvements, but one of the 1986. Dynamic optimalization of CO2 con- Smalltalk-80, the language and its imple-
limiting factors has been the cost of centration in relation to climate control in mentation. Addison-Wesley. Reading,
developing, validating, and implement- greenhouses, p. 147–160. In: H.Z. Enoch Mass.
and B.A. Kimball (eds.). Carbon dioxide
ing the required computer programs, enrichment of greenhouse crops. CRC Heuvelink, E. and H. Challa. 1989. Dy-
models, and knowledge bases. How- Press, Boca Raton, Fla. namic optimization of artificial lighting in
ever, because software development greenhouses. Acta Hort. 260:401–412.
tools are getting more and more pow- Challa, H., G.P.A. Bot, E.M. Nederhoff,
and N.J. van de Braak. 1988. Greenhouse Hooper, A.W. 1988. Computer control of
erful, we can envisage a rapid increase the environment in greenhouses. Comput.
in the use of “knowledge-intensive” climate control in the nineties. Acta Hort.
230:459–470. Electron. Agr. 3:11–27.
environment control systems.
Gauthier, L. 1993a. GX: A Smalltalk-based Meyer, B. 1988. Object-oriented software
platform for greenhouse environment con- construction. Prentice-Hall, New York.
Literature Cited trol. Part I—Modelling and managing the Spitters C.J.T., H. van Keulen, and D.W.G.
Bakker, J.C., L. Van den Bos, A.J. Arendzen, physical system. Trans ASAE 35:2003– van Kraalingen. 1989. A simple and uni-
and L. Spaans. 1988. A distributed system 2009. versal crop growth simulator: SUCROS87.
for glasshouse climate control, data acqui- Gauthier, L. 1993b. GX: A Smalltalk-based In: R. Rabbinge, S.A. Ward, and H.H. van
sition and analysis. Computer Electron. platform for greenhouse environment con- Laar (eds.). Simulation and systems man-
Agr. 3: 1–9. trol. Part II—Supporting and implement- agement in crop protection. PUDOC.
ing control strategies. Trans. ASAE 35: Wageningen. Netherlands.
Brault, D., R. Boily, and G. Gueymard.
1989. Contributions of HPS lighting to 2011–2020. Udink ten Cate, A.J. 1980. Remarks on
the heating requirements of a greenhouse. Gauthier, L. 1993c. An object-oriented greenhouse climate control models. Acta
Amer. Soc. Agr. Eng., St. Joseph, Mich. greenhouse simulation software system— Hort. 106:43–48.
ASAE Paper 89-4039. Part I: Architecture and functional descrip- Udink ten Cate, A.J., G.P.A. Bot, and J.J.
Bruggink, G.T. and E. Heuvelink. 1987. tion. Can. Agr. Eng. 35:215–222. van Dixhoorn. 1978. Computer control of
Influence of light on the growth of young Gauthier, L. 1993d. An object-oriented greenhouse climates. Acta Hort. 87:265–
tomato, cucumber and sweet pepper plants greenhouse simulation software system— 272.
in the greenhouse: Calculating the effect of Part II: Description, validation and use of Vézina, F., M.J. Trudel, and A. Gosselin.
differences in light integral. Scientia Hort. a simple model. Can. Agr. Eng. 35:223– 1991. Influence du mode d’utilisation de
31:175–183. 228. l’éclairage d’appoint sur la productivité et
Carrier, M., L. Gauthier, and A. Gosselin. Gauthier, L. and R. Guay. 1990. An ob- la physiologie de la tomate de serre. Can. J.
1993. Adaptation of the SUCROS87 crop ject-oriented design for a greenhouse cli- Plant Sci. 71: 923–932.
growth model to the use of supplemental mate control system. Trans. ASAE 33:999–
lighting in greenhouses. Scientia Hort. (In 1004.
press.)