Energy Management Optimization Of Integrated ... - IEEE Xplore

Proceedingsof the 1998 IEEE International Conference on Control Applications Trieste, Italy 1-4 September 1998

TPO4

Energy Management Optimization of Integrated Generation Systems by Fuzzy Logic Control Francesco Bonanno Giovanni Patani Department of Electronic Engineering and Application of Mathematics, University of Reggio Calabria

Via Graziella, Feo di Vito, 89060 Reggio Calabria, Italy e-mail fbonanno@drpmat unict

it

Institute of Machines, University of Catania Viale A. Dona, 6 , 95125 Catania, Italy Phone + 39 95 330240 Fax + 39 95 330258

The typical applications , such as small isolated communities, show strongly irregular load diagram during the time of day and the weeks too and the use of renewable sources, according to their availability, represents a constraint either for the conventional generating group and for the energy storage system. Constraints rise due to the need to supply the power load demand dowing also a regular and efficient diesel group operation in terms of start-stop cycle, even in low renewable energy sources conditions. Several papers appeared in Iicterature about fiizzy logic application in power system [4],[5],[6],[7]but its application in IGS was not extensively investigated. Initial works on this subject have been proposed by the authors in [SI. In [9] Horodecki et uf discusses about three methods for the technical and economical assessment of wind-photovoltaic battery systems by fuzzy logic but no pratical application, no case study was enclosed in the paper In [IO] Kothari et a1 develops a hzzy dynamic programming for optimal generator maintenance scheduling incorporating load forecasting. The use of hzzy sets in conjuction with dynamic programming is valuable because provide an aid to the power systems manager in handling factors and ucertainties of the maintenance scheduling. In IGS the nature of diesel generator maintenance scheduling problem is more and more complex due the enclosed generating units to develop power by renewable sources and batteries storage The long term operating of IGSs [ 1 I] show as the diesel groups often works at low regime and depending on the meteorological conditions, that in tum affect wind speed and solar radiation values, their on-off cycles is changed by the plant operator aiming to oil saving so increasing the number of diesel units start-stop. In IGS the aim of maintenance scheduling at minimum costs, to meet the load demand in order to assure high reliability plant operating is constrained to the renewable energy production too. Accordingly wind speed and solar radiation data forecasting must be carellly performed because experimental data of the potential installation site are not often available. The load, wind, solar forecasting

Abstract.- Energy Flows Manag.&ent (EFM)is very important task for Integrated Generation Systems (IGSs) designers and planners because in practice the amount of renewable source for long time operating is unknown depending on meteorological conditions. The wide variety of renewable energy sourca and their highly site-specitic and variable nature coupled with different types of load demand claim for efficient and reliable EFM of IGSc The forecasting errors of the meteorological data and model calculated developed powers, so as in power load demand have to be accounted to avoid poor perfomance. Fuzzy sets enable to handle these uncertainties and so making reliable and efficient EFM of IGS. In this paper B fuzzy logic approach has been used to address the EFM problems in IGS. As a case of study a proposed rule based EFM strategy has been implemented to a small sized IGS experimental prototp assembled in the h e of the project of the European Union “Combined multiple renewable energy source systems simulator”. The enclosed results show relevant improvement in e n e r ~perormance, so as problems that rise by fuzzy logic application in IGS. The comparison of the simulation results that is done by an in house developed logistical program embedding conventionalEFM point out as the Fuzzy based EFM reduces fuel consumption and increase the penetration of renewable sources. 1.

,

IGSs Energy Management problem area

Integrated Generation Systems (IGSs) are increasingly considered in order to produce more electrical energy by renewable sources in isolated area, small islands, rural communities or less developed countries[ I]. Different solutions can be proposed in order to integrate conventional and renewable energy sources and several kind of generating units as photovoltaic, wind, diesel, and battery storage have to be assembled to built up an Integrated Generation System(IGS). These groups have different characteristics in delivering energy [2],[3] and some problems must be faced due their contemporary connection to supply load. The main advantage of an IGS versus conventional diesel station alone is the he1 saving with less diesel groups maintenance as a results of the integration of the renewables sources. It should be noted that without renewables integration the diesel groups siting has to be made to meet the peak power demand.

0-7803-4104-x/98/$l0.00 0 1998 IEEE

969

2. Management and control of

errors can be relevant over long total time horizon Moreover the value of average forecast error and the percentage error in wind, solar and load forecasting should be selected by their statistical data evaluation for several past years Accurate modeling of these data is so needed and further efforts have to be made In less developing countries, small island and in may potential installation sites data are not available because a few of IGS are worldwide operating On the contrary in IGS the main advantages of using fiizzy logic lie in the EFM problem area In fact the management of the enclosed generating groups according a proper diesel strategy constrained by the manufacturers data about optimal on-off cycle, must take in account the of predicted wind and photovoltaic power and load demand An intelligent EFM assures reduced maintenance scheduling and lead to saving diesel oil [8] The present paper describes the Fuzzy Logic application to optmize the EFM of small Sized IGSs and shows its effectiveness to exploit the availability of qualitative information about optimal IGSs operating of the conventional and renewable generating units as well as IGSs as a whole coming &om manufacturers and experience of operating Fuzzy logic or computing with words as stated by Zadeh [ 121 will became a necessity in EEM problems of IGS being the evaluation of renewable imprecise and moreover the tolerance for imprecision can be exploited to achieve robustness in the real IGS operating Exploitation of the tolerance for imprecision in a real world problem as IGSs EFM has been the motivation for the fUzzy logic application The EFM described in natural language is implemented by 130 fuzzy IF-” rules In order to include fuzzy sets for renewables, load demand, diesel groups power and discharged amperhour by the batteries and rule based EFM for IGS long term simulation of operating FuzzyTechQ software was used[ 131 This software provide p o w e h l GUI and the $-link statements allows the communication between the CC+ application enclosing the lead acid charge-discharge battery model and the wnd turbine and photovoltaic inverter system models The IGS prototype enclose two diesel groups, with two engines and two synchronous generator rated respectively at 30 and 50 kVA, a wind generation group, including turbine and generator rated at 50kW, a photovoltaic field simulator rated at 29 kWp, a dump load and a batteries energy storage of 135kWh(ii300V Testing of intelligent EFM application on larger sized IGSs will not be provided being the typical power rated for several plant in the range 70- 15OkVA

IGSS

IGSs are dynamic systems and the related control systems and regulators have to be designed as reported in [14] but the problem of IGSs control is also one of lGSs EFM. In fact at the design stage of IGS generally average load demand and renewable and conventional supply pattem are used , while during their operation, fluctuation in wind speed solar radiation and load demand will occurs. The EFM by an intelligent controller is essential to estabilish how much energy is available and how much has to be used by diesel sets to supply load. In general a three level fiinction is required to the controller of an IGS [ 3 ] The first is the standard action of regulation performed on both the voltage and speed of each generator with the aim to minimize bus bars voltage and fiequency deviations. The second level of control concerns the actions performed on the load demand introducing priority techmques or the influence of external factors such as price variations, wheather conditions, demand side hystograms in order to optimize the energy conversion process in every conventional and renewable energy subsystem. The third level of control consists in a system supervision and EFM aiming to correctly and efficiently share the load demand according to the availability either of renewables, affected by meteorological conditions, and diesel groups affected by on-off cycle. Such control level could be achieved by adaptive technique control which normally needs a considerable amount of informations and long years of experience [15] The next sections will face about the latter level hnction required to the IGSs control system dealing this subject with fiizzy logic methodology Exploiting the knowledge acquired by experiencing a realized facility [2] that can be continuously increased, modified and which must be stored and managed as a whole in order to optmize the IGS performance a rule based model was built up. Little efforts have been devoted in the area of intelligent decision making approaches to optimize the management of the IGS so as to provide their automation while rule-based approaches are suited for decision making task [8]. The reasons for this was due mainly to the fact that , like any other industries, the IGS field is market led . At present IGSs costs and tecbcal problems are relevant expecially in designing and manufacturing of the power electronics apparatus, as bidirectional self-line commutated inverter , tailored for several configuration [3] The customers are very happy if the use of a small sized IGS can demonstrate a fie1 saving in conjuction with a safe and reliable plant operating. Then the IGS automation problems have been deferred or faced with conventional techniques due the few realized plant.

cess

970

~

3. Fuzzy logic based EFM

This rule based EFM is formulated intuitively according to hardware simulation experience in several configuration about the IGSs prototype [3]. The controller inputs are the powers developed by the generating units supplied by renewables and the power load demand. The fourth input to the fuzzy controller shown in fig 1 labelled as SOC, cannot be strictly regarded as an input being it model calculated depending on power batteries Pbat that is also a f k z y output. During preliminary studies, reported by the authors in initial works , a Fuzzy based EFM was implemented in the manner as follows 0 calculate the variations of the electrical energy to supply the load; 0 divide the energy betwTn conventional and renewable sources according to their availability; 0 divide the energy between wind and photovoltaic sources according to the batteries SOC; 0 divide the renewable and conventional energy surplus between batteries and dump load. According to this proposed control strategy some fuzzy controllers must be employed ; the first one processes the power error signal (difference between total electric power developed by conventional and non conventional sources and power load demand). Based on this error and on its change the first fuuy controller establish the power electric change to develop and the other manage the power flows. This implementation is extremely general, and can be applied to different size of IGSs plant by using variables normalised to the rated total power.

At present time fuzzy logic is a well firmed powerfhl technology that allows to embed qualitative informations, linguistic knowledge and uncertain data in the design of intelligent systems. Different approaches have been presented in licterature to perform a Fuzzy Logic based modelizaton of complex or non-linear plants. The most known are the Rule-Based approach by Zadeh [ 121 and the Relational approach by Sugeno [ 161. Neglecting the economical concerns there are four input fuzzy variables within the selected time windows as shown in figure 2. It is important to note that also the economic considerations have fiury nature. The study is performed by the implementation in Fuuy Tech linked to C t t routine to provide simulation. Triangular shaped membership functions were used for all the inputs feeding the fuzzy controller. In the present case study the values of the diesel power outputs labelled as Pdsl and Pdst were calculated by the centhroid method of dehzzyfication [ 131. The rule base of fig 1 stores 130 linguistic rules of EFM and implemented in F u u y Tech according to the main following points 0 supply load demand (Pcar ) by renewables (Peol,Pphv); 0 diesel operating strategy to reduce the start-stop cycles and to save fuel, evaluate the State of Charghe (SOC); management of the charge-discharge battery operation.

\

f

\ I

I

< MIN

I /

971

/

day to about 7580% SOC while the charge occurs during the off-peak time of IGS operating. The simulation runs starts reading three datafile each containing the hourly mean wind speeds, the hourly global solar radiation values and the hourly electric load demand for the selected site. Then the power developed by renewables are calculated according the model described in the previous section The hourly energy flows are so calculated and stored over the simulation time. The implemented fuzzy controller establishes the power supplied by the enclosed DGs and batteries storage system according to the load demand and keeping in account the SOC. Then dump load energy is wasted according an energy balance equation At the end of the simulation period related to the IGS operating global informations regarding the IGSs EFM as energy available from DGs and renewables, total Oil consumption (OIL), Penetration of Renewable Energy (ERPEL), energy flows of the storage system and SOC, EDL are available for both conventional and fuzzy management. The main outputs simulation are listed in Table I1 and Table I11 The proposed ~~IZZYcontroller provides promising results as shown in the Tables. In these tables each energy is also shown in percentage of the load demand to emphasize the different energy distribution and production by the two different control. The comparison of the simulation results points out as the Fuzzy based EFM reduces fuel consumption, Number of DGs starts-up (WSDG) and increase the ERPEL values The main difference that rises comparing these results is that by the fuzzy based EFM battery storage operate as fuel saver being its use before the second DG Then the global condition of battery storage were at lower SOC value In small islands fuel cost is very high due its trasportation costs The slots of ULE in Table 111 is always empty being the present case study mainly devoted to analyze already sized IGSs according the predicted renewables and load demand Clearly in design optimization studies this issue has to be considered. During winter season the lower renewable production could be lead to not zero values of ULE due the expected more and more low SOC conditions. The use of fuzzy sets for renewables seems very effective also to optimize the EDL values and in turn increasing the ERPEL. The second DG is allowed to be shut down and the &zzy control optimize the oil saving and the number of its starts-up

Labelling as PI and Pet respectively the power load and the total available power the following equation can be written:

or considering these variables with respect to the IGS power rated: e , ( k ) = e ( k ) / Pb (2)

de,, ( k ) = Ae(k) 1 Pb

where k indicate the generic time of sampling. Based on the previous relations the &zzy rules Ri can be written. The main difficulty encountered experiencing this fuzzy EFM, driven by the power error, was the handling of the Diesel Groups @Gs) power level working at minimum allowable loading to assure suitable engines temperature enclosed in the total power Pet . In long term operating these levels of power available according the programmed DGs on-off cycles affect the EFM. This implementation for IGSs with DGs always running, to assure low Unmet Load E nergy (ULE) values leads to poor energetic performance in terms either of fuel saving and wasted Energy in the Dump Load (EDL).

3. Simulation results

The implemented IGSs model as a whole is a quasi steady-state model with one hour time step By using these models the time step range is from 10 minutes to 1hour [ 171. It requires hourly input data of load demand, wind and solar The meteorolgical data were measured in a spring season of a small island in the Mediterraneo sea [ I13 The hourly wind speeds vary between 6 9 and 8 3 d s e s and the solar radiation vary between 0 and 700w/m2 The value of solar radiation have been adjusted to keep in account the tilt angle of the solar array The hourly load demand vary in the range 30-150kW Load profiles on long term operating have been generated by powerful statement available in ACSL E181 processing the load pattern available within 24 hours period In the Table I is reported the renewable energy production and the load demand over a typical day, week, twenty days and an entire months of a spring season of potential operating of the sized IGS in small island in the Mediteranneo sea At beginning of simulation the batteries storage is at IOO%SOC value The maximum energy charged or discharged by the batteries storage is taken 10Y0of the sized capacity per hour The battery is discharged each

Table I RENEWABLE ENERGY PRODUCTION AND ENERGY DEMAND

1 SYnulationTime 1 [hour] 24 168

480 744

972

EL [kWh] 1920 14740 43257 64369

I

EW

[kwh] 499.2 3611 3 10425 15448

26% 245% 24 1% 24%

I

EPV [kwh] 230 12% 1872 12 7% 5625 13% 8046 12.5%

1

OIL

ED2

ED I

EDT

Simulation TUTK

Battery Storage Performance Tune [hour1

EBAT drschorge

EBAT charge

EDL

[kwh]

Fviil]

prwh]

NOS

NOS

DG1

DG2

Table III Conventional (C) and Fuzzy'based control (F) ERPEL

SOC

ULE

YO

24

C

99

052%

-768

04%

291

152%

2286

098

F

384

-768

04%

61

60

04%

197 1991

1026% 135%

277

C

2% 041%

081 097

237

performance. In general similar calculations performed in logistical energetic performance prevision of IGS refers to spring season so leading in renewables overestimation over the other operating period. Sometimes these datafile processing led to failures of fuzzy based EFM over long term operating: high ULE value, unexpected energy surplus fed to the batteries, unexpected diesel oil saving and diesel energy wasted by dump load. Generally remote areas or less developed countries where renewable energy can make their greatest impact have not stored experimental data on the available renewable sources. In this case the designer has to estimate the resource based on meteorological data at similar or nearby sites, or by exploitation of the qualitative information coming from discussion with local people. These latter information are in vague form: mostly cloudy, highly windy. Moreover discussion with local people has to be made in order to estimate the daily energy need. Imprecision is intrinsic in the use of words. The installation on site of meteorological station provides only a short term information about wind and sun conditions and not The key point is represent long term averages. incorporating hzzy decisions for EFM of IGS so accounting the randomly variation of the renewable and load demand. Then a hzzy logic control strategy which can

In the present study the start up time of DGs is relatively small thus the second DG , that in general should be kept as power reserve, is shut down if possible. This proposed approach is valuable for already sized IGS, but because the economical concerns have also hzzy nature, their consideration in conjuction with parametric hzzy set for the size of the generating units and storage could be lead to optmized IGS design too It is usehl to point out deficiencies in the design of both generating units and storage system sizing so reducing the LkE. This seems a more effective approach compared with deterministic analysis and linear programming techques used in order to design and analyze IGSs [ 191. These preliminary studies about rule based EFM implementation can provides the basis for hture hardware implementation. 5. Conclusions and discussion The simulation results reported in [8] mainly regard the IGS daily operating in order to emphasize the benefits on the plant automation rised by an intelligent control strategy The fuel saving evaluated by taking in account the typical mean day in a spring season relative to the meteorological condition, and processing the results over a more long time period of operating leads to overestimate the energy

973

Pr"gs

of 30th University Power Engineering Conference UPEC'95, September 5-7. 1995, London, United Kingdom.

effectively combine numerical and linguistic information into IGS to improve its EFM has been proposed. The main aims were saving diesel oil, reducing of the diesel start-stop cycle and battery charge discharge cycling, optimization of the EDL . The main advantages of using fuzzy controller for EFM of IGS are:handling uncertain operating conditions according to the random availability of renewable sources, then robustness in several operating conditions and design by operating experience concerning the plant to be managed. Samples of simulations indicate as Fuzzy based EFM improve the global energetic performance but it's very very difficult design a hzzy controller for the entire year due the highly variable and seasonal nature of the renewable sources and user load demand. The year should be divided into several time segments so avoiding the need of big amount of meteorological data for the fizzy controller input. Then the paper present only the first step of fizzy controller development for EFM problems of IGS. It's only a seasonal design. However it is important for the system planners in TGS area to be aware of the capabilities and limitation of the intelligent control techniques in IGS automation and control because the prediction of the renewable energies plays a fundamental role. The next step in performance evaluation of IGSs managed by &zzy logic control will be incorporating economic performance as generating cost, maintenance and fie1 cost, price and lifetime of DGs and batteries. Fuzzy computing enclosing these figures could be used to compare different IGS configuration. However the importance of these results still holds being the main scope of this research merely to provide a basis for f k z y based EFM of IGSs, SO as to evaluate the improvements in energy performance achieved by fuzzy logic technology application, rather than a commercial enterprice.

[4] R.Perrqaan, TRLunn, " F q Logic Control of Combined Heat and Power System",Proceedmgs of University Power Enpneering Conference WEC'95, September 5-7, 1995, London, United Kingdom, pp. 62 I 624. [5] C G Grc", K W Chan, R W Dunn, A R Daniels, " Funy Logic Techmques Applied to Power System Control", Proceedmgs of Unibersity Power E n g m m g Conference UPEC'94, September 14-16, 1994, Galway, Ireland [6] SF.Noor, J.R.Mcdonald," Incorporating Fuzzy Decisions m Long Term Generation Expansion Planning", Proceedings of Power Engineering Conference UPEC'93, September 6-7, 1993, United Kingdom. [7] Y.H.%ng, R. K. Aggarwal, A.T Johns," Fuzzy Logic In The Development of New Protection Equipment For Power Systems", Proceedings of University Power Engineering Conference UPEC'93. September,1993, United Kingdom

(81 F Bonanno, GCapmi, ''Fuzzy Logic Based Energy Flows Management of Integated Generation Systems", Advances m Intelligent Systems, d t e d by F C Morabito, 1 0 s Press 1997 Senes Front" in Artificial Intelligence [9] A. Horodecki et al , "Advantages of the use of Elements of Fuzzy Logic for the Assessment of Unconventional Power Electronics Jh.;ing S)-stems", Advances in Intelligent Systems, edited by F.C. Morabito, 1 0 s Press 1997 Series Frontiers in Artificial Intelligence. [IO] D P Kothan, A Ahmad, "Fuzzy Uynamc Programng Based Gptmal Generator Mamtenance Schedulmg Incorporatmg Load Forrxlastmg'', Advances m Intelligent Systems , d t e d by F C Morabito m the Senes Frontiers m Artificial Intelligence, IOS Press 1997 [ 1 11 F.Bonanno,A. Consoli, S. Lombardo, A. Raciti, "A Logistical Model for Performance Evaluations of Hybrid Generation Systems". Proceedings of IEEE L-IS Industrial & Commercial Power Sy.ytem Technical

Conference I&CPSPhdadelphia (USA), May 12-15, 1997.

APPENDIX EW total energy from the wind groups Epb'total energ. from the photovoltaic-inverter systems ERP IEL penetration of renewables EDT total energy supplied by the DGs EL total load energy demand ED1 total energy supplied by the first DG ED2 total energy supplied by the second DG EBAT energy fed to the batteries or supplied by the batteries

L. A. Zadeh, "Fuzzy LogicXomputing With Words", IEEE [I21 Transaction on Furzy Systems, vol . 4, NO. 2, May 1996. [ 131 FuzzyTECP" release 4.0, User manual, E'JFORM Gmbh [I41 F. Bonanno, A. Consoli, A. Raciti, B. Morgana, IJ. " m a , "Transient Analysis of Multiple Integrated Generation Systems, DieselWind-PV," accepted for publication on IEEE PES Transaction. [ l j ] L. X. Wang, "Adaptive Fuzzy Systems and Control: Design and Stability Analysis," PTR Prentice Hall, Englewood Cliffs,Ne Jersey, 1994. [I61 T. Takagi, M. Sugeno, "Fuzzy Identification Systems and its Applications to Modeling and Control", IEEE Trans. System Man and Cybernetics,~ o l 15, . pp. 116-132, Jan./Feb., 1985 11 71 D. Child , I. R Smith, "Long Term Performance Modeling of a Combined Energy Gtneration Systtxn", Proceedings of University Power EngineeringConference UPEC'95, September,1995, United Kingdom [IS] Advanced Continuous Simulation Language (ACSL), Reference Manual, Edition 4.2, Mitchell & Gauthier Associates USA 1191 R. Chedid, S Rahman, "Unit Sizing and Control of Hybrid WindSolar Power Systems", IEEE Transaction on Energy Conversion, Vol. 12, No. 1, March 1997.

6. References [ I ] R. Ramakumar, I. Abouzahr, K. Knshnan, K. Ashenaqi, "Design Scenarios for lntegrated Renewable Energy Systems," IEEE Trunsaction on Energy Conversion, Vol. 10, No. 4, December 1995, pp. 736-746. [2] "Combined Multiple Renewable Source System Simulator Facility", Fust and Second Year Report, Work performed under a CEC Contract No. JOU-2CT92-0215,Conphoebus s.c.r.1.. 1994. 131 BOMMO,F., Raciti, A., Testa, A., Leotta, A., Nocem, U., "Analysis of Iiybnd Generation System Cofigurations Utilising Renewable Sources",

974