www.elsevier.com/locate/apthermeng

Applied Thermal Engineering 26 (2006) 2255–2261

The performance of an open-loop lake water heat pumpsystem in south China

Xiao Chen a, Guoqiang Zhang a,*, Jianguo Peng a, Xuanjun Lin b, Tingting Liu a

a College of Civil Engineering, Hunan University, Changsha, Hunan 410082, PR Chinab Hunan Lingtian Co., Ltd., Lingtian Road, Xiangtan, Hunan 411201, PR China

Received 19 April 2005; accepted 18 March 2006Available online 19 May 2006

Abstract

A district heating and cooling (DHC) system that utilizes lake water as heat source–sink of heat pumps has been constructed in Xiangtan,a city in Hunan province in south China. An initial analytical study had been carried out before the construction. In this paper, a simplifiedtwo-dimensional model is developed to simulate the steady lake water temperature (LWT) distribution during continuous operation. Thesimulation results indicate that the impacts of the discharge on entering water temperatures (EWT) and the ecological environment of lakeare acceptable. Field test results showed that the COP values of the system were, respectively, 0.7–0.85 higher in cooling season and about0.46 higher in heating season than those of the air-source heat pump (ASHP) units at the same sink and source temperatures. An acceptablepayback period of 5.6 years was found through an economic analysis based on the comparison between the initial and operating costs of thesystem and those of the distributed ASHP units that would have been installed according to initial scheme.� 2006 Elsevier Ltd. All rights reserved.

Keywords: Heat pump; Lake water temperature; District heating and cooling; Performance

1. Introduction

Ground-source heat pumps, including ground-coupledheat pumps, groundwater heat pumps and surface waterheat pumps (SWHPs) [1], help to reduce energy consump-tion for heating and cooling of buildings and CO2

emissions.Surface water bodies can serve as good heat sources and

sinks if utilized properly [1]. Cantrell and Wepfer studiedthe feasibility of using shallow ponds for dissipation ofbuilding heat in north Ohio [2]. Aittomaki studied the pos-sibility of using lake water as heat source for heat pumpheating in cold climate [3]. Kavanaugh investigated theoperation of water-to-air heat pumps and direct coolingsystem with water of southern lakes in USA [4]. In Turkey,experimental study of Seyhan River and dam lake as heatsource–sink for heat pumps was carried out [5]. The lake-

1359-4311/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.applthermaleng.2006.03.009

* Corresponding author. Tel.: +86 731 8825398; fax: +86 731 8821005.E-mail address: gqzhang@vip.sina.com (G. Zhang).

source cooling project at Cornell University uses the deep,cold water of nearby Cayuga Lake supplying over63,306 kW of cooling for its campus [6].

The application of ASHPs in south China is rather pop-ular. However, the performance and the capacity ofASHPs decrease rapidly with increasing ambient air tem-perature (AAT) during cooling season, and with decreasingAAT during heating season. Some heat pump systems insouth China utilize groundwater as heat source–sink.Though groundwater is a very good heat source–sink forheat pumps, the application of groundwater heat pumpsis restricted by local groundwater conditions and regula-tions. The surface water resources in south China accountfor about 70% of total surface water resources of China [7].However, utilization of surface water as heat source–sinkin heat pump systems is rare in south China.

Xiangtan, with a population of over one million, islocated in the east of Hunan province, China. Accordingto an overall planning made in 2001, City Hall, TV Station,Shop and Grand Theater, would be built around Mengze

Nomenclature

B average width of lake (m)COPc coefficient of performance in cooling modeCOPh coefficient of performance in heating modec specific heat of water (kJ/kg �C)E diffusion coefficient (m2/h)es saturation partial vapor pressure at heated/

cooled water surface temperature (mm Hg)H average depth of lake (m)K heat loss/gain coefficient of heated/cooled water

body (W/m2 �C)L length of every part of the line source (m)mcon flow rate of the water in condenser (kg/s)meva flow rate of the water in evaporator (kg/s)P atmospheric pressure (mm Hg)Q0 discharge volume flux (m3/h)T0 discharge temperature (�C)Ta ambient water temperature (�C)Tci inlet water temperature of the condenser (�C)Tco outlet water temperature of the condenser (�C)Tei inlet water temperature of the evaporator (�C)Teo outlet water temperature of the evaporator (�C)Ts heated/cooled water surface temperature (�C)t time (h)

Uw wind speed 2 m above the water surface (m/s)u longitudinal velocity (m/h)Wcom power consumption of the compressor (kW)Wpf power consumption of the lake water pumps

(SWHP) or the fans (ASHP) (kW)

Greek symbols

e emissivity of waterh temperature difference between heated/cooled

water and ambient water (�C)hc centerline temperature difference (�C)q0 discharge water density (kg/m3)qa ambient water density (kg/m3)qc centerline water density (kg/m3)r Stefan–Boltzmann constant, 5.67 · 10�8 W/

m2 K4

s time-step (h)

Subscripts

0 dischargea ambient waterc centerline

2256 X. Chen et al. / Applied Thermal Engineering 26 (2006) 2255–2261

Lake, which is 67,000 m2 in area and 3 m in average depth.Fig. 1 shows daily average AAT and daily LWT 1 m belowthe surface at 9:00 AM in 2001 obtained from local weatherbureau and hydrological station, respectively. The LWTwas 2–5 �C lower/higher than daily average AAT duringmost time of cooling/heating seasons. Sometimes the differ-ence between the two was as high as 8 �C. The favorableLWT make Mengze Lake a good heat source–sink alterna-tive to air for heat pumps. Thus, the investors of fourbuildings were inclined to build a lake water heat pump

Fig. 1. Daily average AAT and LWT at 9:00 AM in 2001.

system for DHC instead of initial scheme that ASHP unitswould have been installed. The system has been operatingsince summer 2003. This paper presents the system intro-duction, the initial study and its practical performancebased on field tests.

2. System description

2.1. System load

A program named HDY-SMAD [8] was used to predictthe hourly cooling load in a cooling day and the hourlyheating load in a heating day under respective design con-ditions. Table 1 gives gross floor area (GFA), peak coolingload and heating load per unit GFA (CLU and HLU),peak cooling load and heating load of four buildings.The peak cooling and heating loads of the plant were foundto be 12,196 kW and 6953 kW, respectively. The diversityfactors in cooling and heating seasons were 0.82 and 0.86respectively.

2.2. System configuration

A simplified system schematic is shown in Fig. 2. Theplant is located under a public square, which is nearly thecenter of the four buildings. The water intake 110 m awayfrom the plant is located at the depth of 2 m from the watersurface. Eight valves are employed for the switch between

Table 1List of GFA and peak cooling and heating load data

Building GFA(103 · m2)

Peak CLU(W/m2)

Peak HLU(W/m2)

Peak cooling load(kW)

Peak heating load(kW)

City Hall 28.1 132 98 3696 2754TV Station 13.4 231 117 3094 1568Shop 22.3 210 106 4686 2364Grand Theater 11.6 292 124 3387 1438

Total 75.4 14,863 8124

Fig. 2. DHC system schematic with lake water heat pumps.

Fig. 3. Schematic of Mengze Lake.

X. Chen et al. / Applied Thermal Engineering 26 (2006) 2255–2261 2257

the cooling and heating modes of 14 screw water sourceheat pump (WSHP) units in the plant. Lake water ispumped through the water cleaners and the heat pumpsby the lake water pumps installed in the plant. The tai-lor-made water cleaners can remove suspending particlesand floating alga in lake water without use of any chemi-cals. Such cleaning method does not cause harm to thewater environment of Mengze Lake. For the purpose ofsaving pumping energy, a distributed pumping system isintroduced. The constant speed primary pumps areinstalled in the plant, and the variable speed secondarypumps are located in every building.

The design temperature differences of lake water in cool-ing and heating modes are 8 �C and 5 �C respectively. If theevaporator leaving temperature drops to 4 �C with thedecrease in EWT in heating mode, the temperature of evap-orator surface will possibly be lower than 0 �C, which willcause ice accumulation, and make the WSHP units shutdown automatically. Measures must be taken to maintainthe evaporator leaving temperature higher than 4 �C.When the EWT drops near 9 �C, the operators begin toincrease the lake water flow rate appropriately to lowerthe temperature difference of lake water until it reachesthe minimum value of 3 �C. When the EWT drops near7 �C, the two-way control valve will be used to regulateappropriate steam flow rate. Part of the entering lake waterflows into the heat exchanger, and is heated by the steam.Even though the EWT gets lower, the evaporator enteringtemperature can be maintained above 7 �C due to the con-tribution of auxiliary heat.

3. A simplified thermal model for Mengze lake used

as heat source–sink

Several models have been developed for simulation ofcooling ponds or lakes used as heat source–sink. The mod-els developed by Cantrell and Wepfer [2] and Chiassonet al. [9] assumed the entire water body to be at the sametemperature. Pezent and Kavanaugh [10] developed aone-dimensional model for lakes used with WSHPs, whichcontained three energy balance equations for the upperconvective zone, the non-convective zone and the lowerconvective zone, respectively. The model was more suitablefor a close-loop system of a stratified deep lake than for anopen-loop system of a non-stratified shallow lake. To pre-dict the influence of thermal discharge of power plants,two-dimensional and three-dimensional models were devel-oped [11,12]. The models contained continuity, momentumand transport equations, and needed to be solved by time-consuming finite element method or finite volume method.In this study, a simplified two-dimensional model is devel-oped for the case with relatively small discharge flux.

When the operation begins, the heated or cooled wateris jetted into still water perpendicularly (Fig. 3). For theheated water in cooling season, the near-field of jet is com-prised of two regimes: an initial regime of strong jet mixingin the vertical and horizontal directions, followed by aregime of increased buoyancy induced spreading [13].Beyond the transition distance, a pool of buoyant effluentspreads in all directions, and the centerline dilution in thefar-field region will reach stable state [13]. In heating sea-son, it is reasonable to consider the jet as the one jetted intothe water with the same density approximately, for a tem-perature difference of 2–5 �C at low water temperaturesresults in a very little density difference. The centerline

2258 X. Chen et al. / Applied Thermal Engineering 26 (2006) 2255–2261

dilution of the jets in cooling and heating seasons will besteady after a period of time. Eventually, the diffusing flowalong the X-direction will form, and the jet/plume will bedeflected weakly.

To carry out the analysis in the downstream region,following assumptions are made:

1. The transverse velocity can be neglected except for a cer-tain area around the intake, and the longitudinal veloc-ity is regarded as constant, i.e. u = Q0/(BH).

2. Advection dominates the longitudinal transport process,and the longitudinal dispersion can be neglected due tothe small velocity and the shallow water.

3. The diffusion coefficients are isotropic, i.e. Exx =Eyy = E.

4. The bottom and bank of lake are adiabatic.

The vertically averaged two-dimensional transportequation is employed in the following way:

ohotþ u

ohox¼ E

o2hox2þ o2h

oy2

� �� 3:6Kh

qacHð1Þ

K is heat loss/gain coefficient of heated/cooled water, andcan be calculated by [14]

K ¼ oes

oT s

þ PCb

� �ð9:2þ 0:46U 2

wÞ þ 4erðT s þ 273Þ3 ð2Þ

es ¼ exp 20:85� 5278

T s þ 273:3

� �ð3Þ

where Cb is a constant and usually equal to 6.1 ·10�4 �C�1.

Regarding the centerline of the zone of established flow ofjet/plume as a line source, we can solve Eq. (1) analytically.In this study, CORMIX3 model incorporated in an expertsystem named CORMIX for surface and submerged jetsbased on experimental measurements [15] is used to calculatethe centerline dilution of jet/plume. CORMIX3 has beenapplied in the near-field of surface discharge previously byseveral researchers [12,16,17]. The width and length of thedownstream region are divided into n and m equal parts,respectively. The temperature rise or fall of a node in atime-step resulting from the line source after the time ofx/u will be

h0 ¼ nsqaB

Xn

j¼1

Ljhjcq

jcffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

4pEx=up X1

s¼�1exp �

uðy � 2sB� yjÞ2

4Ex

" #((

þ exp �uðy � 2sBþ yjÞ

2

4Ex

" #))ð4Þ

Table 2Parameters for the simulation

Daily averageload (kW)

Number of on-lineunits

Flow(m3/

Cooling mode 9560 11 1320Heating mode 5205 6 792

In practical application, a relatively accurate value can beobtained by letting s = 0, ±1. The temperature of a nodeafter the ith time-step is

hi ¼ hi�1 þ h0 � 3:6Kiðhi�1 þ h0ÞsqacH

ð5Þ

Near the intake, the impacts of the transverse velocity onthe EWT are taken into account. The EWT is calculatedbased on heat and mass balance of a certain area wherewater converges at the intake.

4. Simulation results

According to the proposed scheme, the outlet was 300 maway from the intake and 10 m away from the bank(Fig. 3). To predict the impacts of the discharge on EWTsand on the ecological environment of Mengze Lake, theLWT distribution during continuous operation was simu-lated. It was assumed that the plant operated at its dailyaverage load, and the lake water flow rate, the dischargetemperature and the ambient water temperature (AWT)were all constant. The average wind speed was 2.4 m/s,and the time-step was 0.05 h. The parameters for the simu-lation are given in Table 2.

Fig. 4(a) and (b) shows the LWT variations at differentcross sections, respectively, in cooling and heating modeswhen the LWT distribution was stable. The jet was dilutedrapidly in the near-field due to the horizontal and verticalentrainment processes until the centerline dilution tendedto be stable in the far-field. In cooling mode, the LWT dis-tribution reached stable state 65 h later, and the stableEWT was 30.8 �C. The condensation heat was lost at theair–water interface. It can be concluded from Fig. 4(a) thatthe local LWT rise in the area of 160 · 20 m2 (shown inFig. 3 using dashed lines) in the downstream regionexceeded 2 �C. Considering the local LWT rise in theupstream region would exceed 2 �C, the thermal dischargemight bring some adverse impacts on local ecological envi-ronment of Mengze Lake. But the impacts on the wholeecological environment would be acceptable. In heatingmode, the LWT distribution was steady 106 h later, andthe steady EWT was 8.4 �C. The cooled water lost less heatat the air–water interface than the un-cooled water. There-fore, it could be heated by part of absorbed solar shortwave radiation and atmospheric long wave radiation.When the LWT distribution reaches stable state, the totalrejected/extracted heat is

q ¼ q0cQ0ðT 0 � EWTÞ ð6Þ

rateh)

Dischargevelocity (m/s)

Dischargetemperature (�C)

AWT (�C)

1 37.3 29.50.6 4.9 9.2

Fig. 4. LWT distribution at different cross sections: (a) in cooling modeand (b) in heating mode (B = 100 m).

Fig. 5. Daily (a) maximum EWT during cooling season and (b) minimumEWT during heating season.

X. Chen et al. / Applied Thermal Engineering 26 (2006) 2255–2261 2259

When the EWT was steady during continuous opera-tion, the total rejected/extracted heat would not meet thepeak loads imposed by the plant in cooling/heating modes.To avoid high EWT in summer, the fountains near theintake should be run at peak climatic or cooling loads toreject more heat through spray cooling. Auxiliary heatshould be added in some cold days due to the droppingEWT. In practice, system operation would stop at nightfrom 24:00 to 8:00, and the loads on Saturday and Sundaywould decrease remarkably. The LWT could be recoveredto a certain extend for above reasons.

We simulated the LWT distribution at different AWT.When the AWT decreased or increased by 2–3 �C on thepremise that the discharge temperature difference wasunchanged, centerline temperature difference of the jet/plume would change slightly, but the temperature differ-ence between the EWT and the AWT was nearlyunchanged. The variation of the EWT was nearly equalto that of the AWT. The heat loss/gain process in thedownstream region could counteract the effect of slightchange in centerline temperature difference to a largeextent.

5. Field study of the system performance and comparison

with ASHP performance

Field tests were performed from June 2003 to June 2004to acquire practical data about system operation. Fig. 5(a)and (b) shows, respectively, daily maximum EWT duringthe cooling season in 2003 and daily minimum EWT fromNovember 16, 2003 to March 15, 2004. Like the daily LWTshown in Fig. 1, the daily maximum/minimum EWT dur-ing the cooling/heating seasons showed obvious seasonal-ity. The range of daily maximum and minimum EWT gotwider than that of daily LWT at 9:00 AM in 2001. Themaximum EWT in cooling season was 31.6 �C. In heatingseason, heat extraction and increasingly cold weather madethe EWT drop. The days in which daily minimum EWTwas below 7 �C totaled 17.

Cooling-dominated buildings reject more heat than theyextract annually. For a SWHP application, annually unbal-anced heat rejection and extraction do not cause surfacewater temperature higher and higher in following yearsdue to the heat exchange at the air–water interface unlessabnormal weather conditions. In our case, the LWT hadbeen recovered to normal level before the following coolingseason began. The result of the LWT measurementsshowed that the LWT at 9:00 AM on June 1, 2004 was24.6 �C, which approached the LWT on June 1, 2003.

Fig. 6. COP values of the WSHP unit, the SWHP system, and the ASHPunit at (a) various sink temperatures and (b) various source temperatures.

2260 X. Chen et al. / Applied Thermal Engineering 26 (2006) 2255–2261

A part of this study was to compare SWHP performancewith ASHP performance in south China. COP values of theSWHP system, the WSHP unit and a screw ASHP unit in abuilding equipped with ASHP units in Xiangtan underdifferent operating conditions were determined. The mea-surements in heating season were performed under theoperating conditions of no auxiliary heat provided for lakewater and for defrosting, so that COP values of the SWHPsystem and the ASHP unit could be evaluated from the fol-lowing equations:

COPc ¼mevacðT ei � T eoÞ

W com þ W pf

ð7Þ

COPh ¼mconcðT o � T ciÞ

W com þ W pf

ð8Þ

Chilled/hot water temperatures, EWT and AAT weremeasured by T-type thermocouples with an uncertaintyof ±0.2 �C. Water flow rate and power consumption weremeasured, respectively, by ultrasonic flowmeters with anuncertainty of ±1.5% of actual flow rate and by digitalpower meters with an uncertainty of ±1% of reading.The uncertainty associated with COP was calculated basedon the root-sum square formula given in Ref. [18]. Themaximum uncertainty of COP would be 6.3% forthe WSHP unit, 6.1% for the SWHP system and 6.7%for the ASHP unit. Fig. 6(a) and (b) shows COP valuesof the SWHP system and the WSHP unit, respectively, incooling and heating seasons at various EWTs. Also shownin two figures are COP values of the ASHP unit at differentAATs. These values were determined under the operatingconditions of capacity at 80–90% of full load and chilled/hot water supply temperatures near 7 �C/45 �C.

As the power consumption of lake water pumps wastaken into account, COP values of the system were 9.4–13% lower than those of the WSHP unit. The power con-sumption for pumping surface water brings considerableinfluence on the performance of a SWHP system. Neededsurface water flow rate, distance and elevation betweenwater source and a building are all the considerations fordetermining the suitability of using surface water as heatsource–sink.

Through a linear curve fit to the data points, COP valuesof the SWHP system and the ASHP unit could beexpressed as the following equations:

COPSWHPc ¼ 5:83� 0:066EWT ð9Þ

COPSWHPh ¼ 3:01þ 0:062EWT ð10Þ

COPASHPc ¼ 5:25� 0:071AAT ð11Þ

COPASHPh ¼ 2:42þ 0:073AAT ð12Þ

It can be concluded from above equations that the COPvalues of the SWHP system were, respectively, 0.7–0.85higher in cooling season and about 0.46 higher in heatingseason than those of the ASHP unit at the same sink andsource temperatures. The ranges of source and sink tem-

peratures for the SWHP system were more favorable thanthose for the ASHP unit. Too high or too low AATs led torapid decrease in the ASHP performance. ApplyingSWHPs for DHC in our case can eliminate some defectsof ASHPs such as defrosting in cold weather, low COPand reduced indoor comfort conditions at unfavorableAATs.

6. Economic analysis

The economic analysis for providing a central plant withlake water heat pumps was performed according to localeconomics. The analysis compared the initial and operatingcosts of the SWHP system with those of the distributedASHP units that would have been installed according toinitial scheme.

The initial cost of the WSHP units is nearly equal to thatof the ASHP units, for the decrease in installed capacitydue to the load diversity is offset by higher price of aWSHP unit. Annual energy consumption savings were cal-culated approximately by subtracting annual energy con-sumption for distributing chilled/hot water between theplant and the users from annual energy consumption sav-ings resulting from the higher COP of the SWHP system.Average COP values of the SWHP system and the ASHP

Table 3Economic parameters and comparison

1. Economic parametersDiscount rate 5.58%Inflation rate 3%Electric utility rate $0.092/kW hLife cycle 20 years

2. Economic comparisonOperating cost savings Utility cost savings $90,220/year

Maintenance cost savings $1100/yearAdditional investments Distribution system $410,000

Lake water system $63,000Payback period 5.6 years

X. Chen et al. / Applied Thermal Engineering 26 (2006) 2255–2261 2261

units in every month of cooling and heating seasons werecalculated using Eqs. (9)–(12) in terms of average EWTand AAT in every month. Table 3 shows the economicparameters and the comparison. The payback period wasfound to be 5.6 years, which was acceptable for theinvestors.

7. Conclusions

The investigation of the key technical problems beforeapplying lake water heat pumps for DHC has been intro-duced. A simplified two-dimensional LWT model has beenpresented. The impacts of the discharge on EWTs and theecological environment of Mengze Lake were evaluatedthrough simulation. The field test results showed that theEWT was favorable in most time of cooling and heatingseasons. The SWHP system could provide the buildingswith enough heat in heating season except that auxiliaryheat should be provided in some cold days. Pumping powerrequirement should be considered seriously for a feasibilitystudy of using surface water as heat source–sink. The eco-nomic analysis indicates a relatively short payback periodof 5.6 years.

Acknowledgements

This work was financially supported by the Teachingand Research Award Program for Outstanding YoungTeachers in Higher Education Institutions of MOE, PRChina and Hunan Lingtian Co., Ltd.

References

[1] ASHRAE Handbook: Applications, American Society of Heating,Refrigerating and Air-conditioning Engineers Inc, Atlanta, GA, 1999.

[2] J.M. Cantrell, W.J. Wepfer, Shallow ponds for dissipation of buildingheat: a case study, ASHRAE Transactions Part 1 90 (1984) 239–246.

[3] A. Aittomaki, Lakes as a heat source in cold climate, in: Proceedingsof International Congress of Refrigeration, Washington, USA, 2003.

[4] S.P. Kavanaugh, M.C. Pezent, Lake water applications of water-to-air heat pumps, ASHRAE Transactions Part 1 96 (1990) 813–820.

[5] O. Buyukalaca, F. Ekinci, T. Yilmaz, Experimental investigation ofSeyhan River and dam lake as heat source–sink for a heat pump,Energy 28 (2003) 157–169.

[6] P. Tim, W.S. Joyce, Lake-source cooling, ASHRAE Journal 44 (4)(2002) 37–39.

[7] Ministry of Water Resources of PR China, Water Resources Bulletin,the People’s Republic of China 2002, Beijing, China, 2003. Availablefrom: <http://www.mwr.gov.cn/english>.

[8] HDY-SMAD: An Air-Conditioning Load Calculation and AnalysisProgram, Shanghai HDY Information Technology Co., Ltd., Shang-hai, China, 2001.

[9] A.D. Chiasson, J.D. Spitler, S.J. Rees, M.D. Smith, A model forsimulating the performance of a shallow pond as a supplemental heatrejecter with closed-loop ground source heat pump systems, ASH-RAE Transactions Part 2 106 (2000) 107–121.

[10] M.C. Pezent, S.P. Kavanaugh, Development and verification of athermal model of lakes used with water source heat pumps, ASHRAETransactions Part 1 96 (1990) 574–582.

[11] J.M. Hamrick, W.B. Mills, Analysis of water temperatures inConowingo Pond as influenced by the Peach Bottom atomic powerplant thermal discharge, Environmental Science & Policy 3 (2000)197–209.

[12] S.W. Suh, A hybrid near-field/far-field thermal discharge model forcoastal areas, Marine Pollution Bulletin 43 (2001) 225–233.

[13] G.H. Jirka, E.E. Adams, K.D. Stolzenbach, Buoyant surface jets,Journal of the Hydraulic Division, ASCE 107 (1981) 1467–1487.

[14] V.Z. Antonopoulos, S.K. Gianniou, Simulation of water temperatureand dissolved oxygen distribution in Lake Vegoritis, Greece, Ecolog-ical Modelling 160 (2003) 39–53.

[15] G.H. Jirka, R.L. Doneker, S.W. Hinton, User’s Manual for COR-MIX: A Hydrodynamic Mixing Zone Model and Decision SupportSystem for Pollutant Discharges into Surface Waters, US Environ-mental Protection Agency, Env. Research Laboratory, Athens,Georgia, 1996.

[16] C. Valeo, I.K. Tsanis, H. Shen, Modelling Mimico Creek as a surfacedischarge, Journal of Hydraulic Research 34 (1) (1996) 115–131.

[17] M.J. Davison, K.L. Pun, Hybrid model for prediction of initialdilutions from outfall discharges, Journal of Hydraulic Engineering124 (12) (1998) 1188–1197.

[18] S.J. Treado, T. Snouer, Measurement considerations for the deter-mination of central plant efficiency, ASHRAE Transactions Part 1107 (2001) 401–406.