SA-96-12-2
`
`Design Issues of Variable Chilled-Water
`Flow Through Chillers
`
`Thomas B. Hartman, P.E.
`,\1ember ASHRAE
`
`ABSTRACT
`
`Variable-speed alternating current (AC) drive technolo(cid:173)
`gies are of particular interes!for heating, ventilating, and air(cid:173)
`conditioning (H VA C) designs because controlling motor
`speed with variable-frequency AC drives to achieveflow mod(cid:173)
`ulation provides an opportunity to capture exceptionally high
`part-load operating efficiencies. Since HVAC systems spend
`long hours operating at part-load conditions, improvement in
`part-load efficiency results in substantial energy savings.
`Applying variable flow to chilled-water systems is particularly
`attractive because chilled-water pumping has !Yro associated
`power costs, directly as pumping power and also as a load on
`the chiller plant. The small temperature differentials associ(cid:173)
`ated with chi/led-wata systems mean load-side stratification
`is generally not a concern. but the small temperature dijferen(cid:173)
`tials do raise concerns about heat transfer at reduced flows.
`
`Presently, typical rariable-flow chilled-water systems are
`designed with flvo chilled-water circuits: primary and second(cid:173)
`ary. The primary circuit is usually a low-head circuit that
`maintains a constant chilled-water flow through the chiller,
`while the secondary chilled-water pump(s) provide variable
`flow to the loads based on their demand for cooling. Because
`the primary circuit is low head and requires relatively low
`power, it is often reasoned that /lVo-circuit configurations
`involve only a small pumping energy penalty and only when
`operating at low loads. However, closer analysis uncovers the
`following true penalties ..
`
`a first-cost penalty for employing two separate pumps,
`
`a part-load chiller efficiency penalty from mixing bypassed
`supply chilled water with the return chilled water, and
`
`a chiller capacity penalty of underutilizing the full chiller
`capacity during high cooling demands at conditions not
`precisely congruent to the design peak conditions.
`
`For many building cooling applications, it is possible to
`design a chilled-water supply and distribution system with only
`a single variablejlow circuit. Such designs can avoid the prob(cid:173)
`lems listed here. However; there are potential pi!falls that must
`be considered before such a system can be successful. This paper
`discusses the benefits and problems associated with a single(cid:173)
`circuit variable-chilled-water-:Ji 'ow system and offers a chiller
`plant control strategy that can provide safe, stable, and reliable
`chiller operation over the entire operating range employed in
`typical HVAC applications.
`
`INTRODUCTION
`
`Traditional chilled-water plant design utilizing variable
`chilled-water flow involves primary/secondary loops with sepa(cid:173)
`rate pumps, as shown in Figure 1. Typically, one low-head
`primary loop pump for each chiller in the primary circuit provides
`aconstant flow through the chiller, while one or more higher head
`variable-flow secondary loop pumps modulate to adjust second(cid:173)
`ary chilled-water Jlow to meet actual cooling demand. The imbal(cid:173)
`ance in Jlow between the primary and secondary circuits results
`in Jlow through the bypass piping circuit. While this configura(cid:173)
`tion satisfies the objective of maintaining constant chilled-water
`flow through thechiller, it may not achieve the highest chiller effi(cid:173)
`ciency at part loads and can limit chiller capacity due to the fact
`that under almost all operating conditions, the flows in the tvvo
`loops are not equal.
`To achieve the full potential of variable chilled-water flow
`in an environment of integrated HV AC equipment operating
`under high-performance control strategies, it is necessary to
`rethink the physical configuration of variable-flow chilled-water
`systems. Recent work (Hartman 1993) has shown that integrated
`control strategies can be employed to operate variable-flow
`chilled-waterdistribution systems at much higher efficiencies by
`coordinating the pump speed directly to the load demands with(cid:173)
`out employing pressure control. It is prudent also to analyze the
`
`Thomas B. Hartman is a principal of The Hartman Company, Marysville, Wash.
`
`Page 1 of 5
`
`GE Exhibit 1007
`
`

`

`1
`-~'---CH_IL_LE_R-1
`
`co"'''"'
`
`Speed
`Pr1mary
`Pump
`
`VO.r"•O.blt>
`Speed
`S.econdory
`Pvnp
`
`LOAD 1
`
`LOAD 2
`
`LOAD 3
`
`Fig11re I Primary/secondary chilled-ux1ter loop.
`
`operation of the primary circuit before a particular configuration
`and control scheme is adopted.
`The striking feature of Figure I is the necessity of having
`two separately powered chilled-water circuits. Designers should
`ask themselves whether this is really necessary. The constant(cid:173)
`flow primary circuit has become accepted design practice
`because it is well known that below certain velocities of flow
`through heat exchangers, a switch to laminar flow may cause
`sudden substantial reductions in heat transfer capacity. The
`purpose of the primary pump is to ensure such a condition never
`troubles the system. However, at very low cooling capacity
`requirements, the heat transfer requirements are also greatly
`reduced, and by monitoring the chiller load, chilled-water
`temperature, and refrigerant temperatures, a properly integrated
`control system can easily adjust the overall system operation if
`water flow becomes too low to provide efficient heat transfer or
`may cause the chiller to approach operating limits If the control
`system is operating with suitable high-performance control algo(cid:173)
`rithms, it can promptly make the necessary corrections to ensure
`efficient and stable operation of the entire system at all load
`conditions. With this in mind, consider the simpler piping
`configuration in Figure 2.
`
`In Figure 2, the chiller itself may be a variable-speed unit,
`but in any case, it is one that offers a high turndown ratio and an
`
`VARIABLE SPEED
`CHILLER
`
`Variable Speed
`Chilled Water
`Pvnp
`
`LOAD 2 1- - &1- . -
`
`~
`I
`
`L 0 A D
`
`. - &1- . -
`
`31
`
`LOAD 4
`
`Figure 2 Single-circuit variable-flow flow chilled-water loop.
`
`increasing coefficient of performance (COP) as the cooling load
`is reduced. The required rate of flow through the chiller depends
`on the cooling load being delivered. This is a good design fit
`because the loads are connected with two-way valves such that
`load-side flow also varies with load. In such a scheme, both the
`chilled-water flow and chiller capacity are adjusted to effectively
`meet all load conditions. A threshold cooling capacity limit is
`defined below which the system does not operate, just as is the
`case with present chiller systems. The potential benefits of a
`single-circuit variable-flow chiller system as shown in Figure 2
`are:
`
`lower tirst cost and lower maintenance costs,
`higher overall chiller plant operating efficiencies, and
`greater flexibility in utilizing full chiller capacity at peak
`conditions.
`
`Before discussing these benefits in detail, let us consider the
`critical issues of such a design.
`
`SINGLE-CIRCUIT VARIABLE-FLOW
`SYSTEM CONSIDERATIONS
`
`Configuring a variable-flow chilled-water system as shown
`in Figure 2 does not mean it will work adequately under all load
`conditions without specific attention to the chilled-water flow
`0\er the wide range of potential operating conditions. To ensure
`effective and efficient operation of the Figure 2 configuration,
`several basic requirements must be met. First, the system must
`not be permitted to operate unless the cooling requirement is
`above a minimum threshold load. The threshold cooling load
`requirement is the lowest stable chiller operating load.
`
`SA-96- 12-2
`
`Page 2 of 5
`
`GE Exhibit 1007
`
`

`

`Next the water flow through the chiller evaporator heat
`exchanger must always be sufficient to maintain evaporator
`temperature within suitable limits. Typically, the chiller manu(cid:173)
`facturer recommends a varying evaporator temperature range as
`a function ofthe chiller load, and the relationship between chiller
`efficiency and e\ aporator temperature is an important consider(cid:173)
`ation as well.
`Finally, for the Figure 2 system configuration to be effec(cid:173)
`tive, the nature of the loads must be such that chilled-water
`temperature can rise as the load decreases. Chilled-water
`systems that require low chilled-water temperature under low(cid:173)
`load (low-flow) conditions must be carefully considered before
`such a configuration is adopted. Examples of systems with such
`special requirements are those that may be called upon to provide
`significant dehumidification at low loads or those supplying a
`variety of loads. a significant number of which may be shut off
`during peak-load conditions.
`In many HVAC applications, these limitations do not pose an
`absolute bmTier to employing the system configuration in Figure
`2. However, a complete analysis under the entire variety of oper(cid:173)
`ating conditions that could be encountered must be accomplished
`to be certain the single-circuit scheme will perfotm satisfactorily
`under all conditions. In typical North American single-building
`applications, even though portions of the building may be unoc(cid:173)
`cupied under certain conditions, the single-circuit configuration
`is usually a good candidate for effective and economical space
`cooling.
`In addition to the load requirements listed above, a single(cid:173)
`circuit variable-flow chiller system will be successful only as
`long as the direct digital control (DOC) system has the capacity
`to integrate the operation ofthe chiller(s), pump(s), and the loads
`the system serves with high-performance control algorithms. The
`start -up sequence for a single-circuit chiller system must provide
`calculations of present and upcoming cooling load requirements
`for all loads served. The cooling system is held off until the sum
`of the calculated load requirements reaches a threshold value that
`depends in part on anticipated upcoming conditions. Once
`enabled, high-performance control algorithms must be employed
`to coordinate the cooling loads, pump speed, and chiller capacity
`to meet the demand for cooling at the loads and meet the opera(cid:173)
`tional constraints of the chiller and other system components.
`
`SINGLE-CIRCUIT VARIABLE-FLOW
`CHILLED-WATER SYSTEM CONTROL
`
`Even if it were feasible, it is not optimal to operate the chiller
`in a single-circuit variable-flow system to maintain a constant
`chilled-water temperature. However, without a clear connection
`between chiller capacity and chilled-water pump flow, control
`can become indeterminate and result in erratic operation as the
`changes in chiller capacity and t1ow affect each other. One of the
`great resistances chiller manufacturers have to varying the flow
`of chilled water through chillers is the ditliculty in establishing
`smooth chiller capacity control under varying flow conditions.
`To control the chiller(s) and pump(s) most effectively, some
`simple mechanism of correlating the operation of the two
`
`together is required. Some approaches have been previously
`discussed (Hartman 1 995). One approach that shows a great deal
`of promise is the use of coordinated chiller and pump (CCP)
`control.
`In coordinated chiller and pump control, both the chilled(cid:173)
`water pump and chiller capacity are controlled to react to
`changes in cooling demand at the loads, so that as loads change,
`pump speed and chiller capacity are adjusted in unison (percent
`chiller electric load is set proportionately to the pump motor
`load). This is a simple and effective way to coordinate the oper(cid:173)
`ation of the chiller and pump. The chiller capacity is adjusted in
`proportion to chilled-water pump power (or to the cube of pump
`speed). In this scheme, at approximately 93% of the design water
`flow through the chiller (and the loads), the chiller is operated at
`80% of maximum electrical demand (0.933). At approximately
`60% of the design maximum chilled-water flow, the chiller is
`operated at approximately 22% of maximum electrical demand
`(the same percentage of maximum electrical demand as required
`by the chilled-vvater pump).
`It is important to note that because the COP of the chiller
`rises as the load decreases, the cooling capacity of the chiller (in
`most typical circumstances) does not fall by the same amount as
`its electrical load reduction (which is controlled to adjust chiller
`capacity). The exact change of chiller capacity with respect to
`electrical draw depends on the type of chiller (variable-speed or
`constant-speed) and evaporator and condenser conditions.
`Furthermore, there may be limits to the capacity adjustment range
`that depend on current operating conditions. For this reason, a
`minimum (and in some cases maximum) evaporator and chilled(cid:173)
`water temperature algorithm and minimum capacity algorithm
`operate in parallel with the direct pump control algorithm as a
`limit to the primary chiller capacity control algorithm.
`CCP control does not directly control chilled-water temper(cid:173)
`ature. However, in regions where it may be required, the CCP
`control strategy may be extended to provide some flexibility in
`adjusting chilled-water temperatures for more or less dehumid(cid:173)
`ification under part-load conditions. In this way, the chilled(cid:173)
`water flow to chiller capacity algorithm can be adjusted slightly,
`depending on humidity conditions, to provide more or less latent
`cooling. A sample algorithm in the operators' control language
`(THC 1988) for a simple CCP control is shown in Figure 3.
`
`BENEFITS OF SINGLE-CIRCUIT VARIABLE-FLOW
`CHILLED-WATER SYSTEMS
`
`Because traditional design has steadfastly adhered to the
`concept of constant chilled-water flow through chillers, the
`industry has never had adequate discussion on the benefits of
`employing variable-flow schemes. Generally,
`it has been
`assumed that the benefits are limited to savings in the cost of the
`primary pump and a small energy reduction. At part-loads,
`however, if one focuses on a comparison between the operation
`of the systems under various conditions, the potential benefits are
`seen to be far more substantial.
`Consider the tlow difference between the primary and
`secondary circuits in a traditional two-circuit system as repre-
`
`SA-96-12-2
`
`3
`
`Page 3 of 5
`
`GE Exhibit 1007
`
`

`

`"'ETEAMINE IF CHI'll...ER PUN1 SHOULD RUN"
`00E\I£RY 1 tr.A
`Q.G..oAAO-ll .. MAX((AHU1MAT·AHU1SAT..sP')•J.oas•AHt.JlFLO'M. 0)
`a.G..oMCH...2 .. W.X{(I\HU2MAT-.AHU1SA.T-SPJ"I.086'"AHU2f1..0W). O)
`O..G..OMO-Ln ... MAX{(AHlX\MAT-#'\Ur\SAT.sPj•l Dae"'AH\JI'IFLOW), 0)
`... ClG...OMC..U .....
`0..G..Q1.l0 - CtG..{)t.U).-U + Cl,G...{)M()....(...2
`U.G-DMD • ClG·DMO • (PROJ-HIGH· TEMP 1 70)
`"STAAT AND SiOPQ;!LLER BAS~D ON OEMAJ-.10
`IF Cl.G.p<J..Nl OFF-FOR 30M AND ClG.OMD > 0 4"CAPACITYTHEN
`STAATD..G--Pt..N-If
`
`END
`IF CLG...f>i...ANT ON ANO CLG...oMO < O.:r CAPACITY OA P'UMP..s.PEED <50
`00 EVAP-TEMP < 3.5 OR CONO-TEMP > M THEN
`STC>' O..G.f>V.Nl
`
`END
`
`'WHEN COOUNG: PlANT IS ON QPE'AATE PVMP ACCORDING TO VA.t..VE PO$ffi0N$'"
`!FOr= CLG-PW<r 00 THEN 1'\JW'-SPEED • ~
`IF ct.G-Pt....ANT ON THEN
`If W.X.VAlVE..POS > ~THEN
`PU~P-sf"EEO ... PUMP-SPEED -t 1
`
`PUMP-SPEED • PUMP-SPEED. {10. AV£-VJ,J...vE.p()S} /10
`
`ELSE
`
`END
`• OI'ERATE CH!Ll.S1 CAPt.CtTY IN ;.ccoAOANC!: l'mH PUMP SPEED
`If Cl().-.PI.).NT ON..fOO < 10M THEN
`Q-lt.R..{)t,(D • 0..4
`
`<>Ul"""D • (>'l)I.IP.sPfi'D/100) ~3
`IF EVAP-TE.IrAP < -40 ~ CHI..A-t:::f..t0/2S THEN C>U.A..ot.ID • CHLR.OOD 10
`IFCONO-TEW' > SOTHENCHl.A-OMO • W.X(CHI.R..,.,.D, "')
`•
`
`END<F
`
`ENDOO
`
`O..G-C>UC-t 1 .. D..G-OMO-ln AAE Thf CUAAENT COOUNG DEMANDS FOR EACH OF THE COOl·
`JN(il.OAO$ SEFivECBYTHECHII.l.ER
`
`PROJ-HIGH-tEMP tS TH£ DAY'S PROJECtED HIGH OUtSIDE TEMP'ERATVRe AS CAl.CVLATEO
`6Y THE DOC SYSTEM
`
`Figure 3 Sample variable-flow single-circuit chilled-
`water plant control program.
`
`sented in Figures 4a and 4b. Figure 4a represents !low at low
`demands for cooling, Under this condition, the tlow in the
`primary loop is substantially higher than that in the secondary.
`The higher temperature chilled water returning from the loads
`mixes with bypassed supply chilled water, which reduces the
`chilled-water inlet temperature, which adversely impacts the
`overall chiller operating efficiency at part-load conditions. To see
`hoi\ important this part-load energy penalty can be, consider
`Figure 5. which shows occurrences of various chiller load condi(cid:173)
`tions in typical oftice buildings in four different regions in North
`America. Notice that for an overwhelming majority of the time.
`the chiller plants in all regions operate at low loads. The right(cid:173)
`hand portion of FigureS shows the annual chiller operating hours
`as a percentage of total building operating hours. Note from
`Figure S that although chillers in warm climates of North Amer(cid:173)
`ica operate longer hours, chiller plants in all climates operate at
`roughly the same overall annual load profile. Note also that the
`North American chiller load profile includes chiller plant opera(cid:173)
`tion at less than 60% of design load form ore than 70% of the
`chiller operating hours. The loss of chiller efficiency because of
`primary/secondary tlow differences at part-load conditions can
`be a substantial penalty in many building HV AC applications.
`
`Now, consider Figure 4b for tlow at high loads. For energy
`efficiency, the tlow of the primary circuit is likely to be less than
`the maximum flow capacity of the chiller. Such a selection
`reduces primary pump horsepower from two perspectives(cid:173)
`lower tlow through the primary circuit and also lower pressure
`drop through the chiller. However, the penalty for less primary
`pump horsepower is that the secondary tlow during peak cooling
`requirements may exceed the primary loop flow. During these
`
`periods. return water fi·om the loads is mixed with chilled-water
`supply. Such conditions occur frequently in chilled-water-to-air
`coils because the load conditions (airflow or air psychrometric
`conditions) do not match the design assumptions precisely.
`With the warmer chilled-water supply, the loads may not be
`satisfied and the chiller may not be capable of operating at full
`load because it cannot compensate below its minimum chilled(cid:173)
`water temperature limit. Such a condition results in chiller
`underutilization. The cooling capacity of the chiller may be
`adequate, but the limitations of the primary circuit do not enable
`the full capacity to be utilized. The only.way to eliminate the
`possibility of this problem in a primary/secondary chilled-water
`distribution system is to increase the flow and pressure capacity
`of the primary pump(s). This then results in an increased energy
`penalty for the overwhelming majority of the hours the plant
`operates below peak capacity.
`In the single-circuit scheme there is no bypass. All return
`chilled water enters the chiller without bypassed supply water.
`Furthennore, the variable-speed pump can be sized for a maxi(cid:173)
`mum flow that is somewhat above the design load assumptions.
`This way the designer is ensured that the full capacity of the
`chiller can be utilized when peak loads occur at conditions that do
`not match design conditions exactly.
`
`SUMMARY AND CONCLUSIONS
`While there are limitations to the employment of single(cid:173)
`circuit variable-flow chilled-water systems for building cooling
`applications, the opportunities available to designers -with the
`expertise necessary to apply integrated high-perfonnance DDC
`to these systems make compelling reasons to consider such
`systems. Single-circuit variable-flow chilled-water systems that
`are carefully designed and operated with integrated high-perfor(cid:173)
`mance controls offer the following:
`
`1. Simpler equipment configurations with accompanying first(cid:173)
`cost savings that can reduce the system costs or be invested
`in higher quality, longer lasting components.
`2. Lower total system energy use than what is possible with
`nonintegrated configurations and control strategies.
`3. Control precision that is superior to that of nonintegrated
`traditional control approaches.
`
`To fully exploit the benefits of emerging integration of
`HV AC equipment with high-pe1formance DDC systems, design(cid:173)
`ers should consider single-circuit variable-flow chilled-water
`systems. For many typical applications, the benefits can be
`substantial.
`
`REFERENCES
`Hartman, T. 1993. Direct digital controls for HVAC systems,
`chapter I. New York: McGraw-Hill.
`Hartman, T. 1995. New horizons for HVAC control. Heating1
`Piping1Air Conditioning, March.
`THC. 1988. Operators' control language: A guide to program(cid:173)
`ming functions for DDC systems. Seattle: The Hartman
`Company.
`
`4
`
`SA-96-12-2
`
`Page 4 of 5
`
`GE Exhibit 1007
`
`

`

`~
`
`C\1 R(Tl.RN
`fROH LOADS
`
`S[C[t.IDARY PuMP rLOV IS L(SS TKA.N PR)HARY ?t..HP f'LOV AT LOV LOAD CONDfT!t:NS...
`AT TH[S:[ T !1-f:S:, A rlOV OF VAT(R EOI.W.. TQ T~ Dtrr(R(NC( IN tt,.O\/S
`'&E TV([N
`P! AND PZ BYPASSES TH( SECCNDARY Cl~UH AND MIXES VlT~ RETuRN VAT(R rRQH
`:,.QAl1S. LDW'ERI~C.. THE R(TURI-l T(l'f>(RA fl..~( TO TH( CHllL(R ti.NO RCOVC!NG TH(
`POSSIBL( Crf"!CICNcY or TH(. CHILl(~.
`
`Figure 4a Chiller flow at load conditions.
`
`tv R( TURN
`TO CHILLER
`
`S(CDNMR'f PVHP
`
`~ v.-RJ>\Bl( r't,.OV
`
`~
`
`PZ
`
`CV SlRPL Y
`TO LQ.i.DS
`
`SECONDAI.?Y f>t)~P rLOV IS C.RfAT(R THAN PRlHARY ?UkP fLO\/ AT P(AK LOAD
`CONDITION$.. AT
`f>'((S( TIHES. A rLO\ol Of VATER (IAJAL lO lHC OlfTER£NCE
`IN fLOVS 3(1\J((N P2 AND Pl f\.OW'S FROM. lHC SECONDARY ROtJRN TO TK(
`HCONDAR'" SuPPLY. RAIS:ING
`lHE SVf'f't,...Y T(]·IP{:RATt.:R( TO LOADS AND
`lHE P(A.X CAPACITY AND EfriCl£1-f:Y OF r;< (NT!kE S'fS.TEt-1
`R(IJUClNG
`
`Figure 4b Chiller flow at peak load conditions.
`
`Figure 5 Office building chiller operation
`
`SA-96-12-2
`
`Page 5 of 5
`
`GE Exhibit 1007
`
`