9th International PHOENICS User Conference Moscow, September 2002

1
9th International PHOENICS User
ConferenceMoscow, September 2002

• A presentation by
• Dr. Paddy Phelps
• on behalf of
• Flowsolve and IAC Ltd

September 2002
2
Predicting air flow and heat transfer in an
anechoic test chamber for industrial chillers
3
Outline of Presentation

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed to date
• Presentation of Results
• Conclusions

4
Chiller Test Chamber Ventilation study

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed to date
• Presentation of Results
• Conclusions

5
Industrial Context

• IAC Ltd design and construct a range of bespoke
  anechoic test chambers.
• Their client in this instance was York Ltd,
  manufacturers of air chiller units for building
  HVAC systems.
• York wish to improve the design of their
  products by testing them at the limits of their
  performance envelope

6
Industrial Context

• Test chamber design brief calls for air supply
  temperatures at chiller intakes to be uniform to
  within 10C .
• Chiller unit intakes are located along upper body
  sides and ends.
• Up to 12 ducted fans on top of unit emit highly
  swirling air extract flow, several degrees
  different from ambient

7
Test Chamber Geometry - 1
8
IAC Ltd Chiller Test Chamber ventilation study

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed to date
• Presentation of Results
• Conclusions

9
Objectives of Study

• Use simulation tools to predict mixing of hot
  swirling extract flow with ambient airflow inside
  test facility
• Provide input to design of chamber air supply /
  extract arrangements, by predicting likely effect
  on airflow patterns
• Confirm client criteria for uniformity of
  temperature at chiller intakes can be met

10
Test Chamber Geometry - 2
11
Test Chamber Geometry - 3
12
Simulation Tool Options

• Direct Experiment
• not applicable - building not yet constructed
• use to confirm other predictive tools
• Wind-Tunnel Modelling
• scale-up and thermal representation difficult
• problem with interpretation of results
• Numerical Simulation
• passive (Gaussian) dispersion models
• Computational Fluid Dynamics

13
IAC Ltd Chiller Test Chamber ventilation study

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed to date
• Presentation of Results
• Conclusions

14
Benefits of CFD Approach

• No scale-up problem
• Three-dimensional, steady or transient
• Interrogatable predictions
• Handles effect of
• blockages in domain
• recirculating flow
• multiple inlets and outlets
• multiple interacting heat sources

15
IAC Ltd Chiller Test Chamber ventilation study

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed to date
• Presentation of Results
• Conclusions

16
Solution Domain(s)

• CHAMBER MODEL
• Solution domain encompasses the test chamber up
  to, but not including, the outlet plenum
• Domain 15.22m by 18.88m by 8m high
• PLENUM MODEL
• Solution domain encompasses the outlet plenum
  only
• Domain 12.2m by 17.08m by 1.3m high

17
CFD Model Description - 1

• Representation of the effects of
• blockage due to the presence of an internal
  obstacle (chiller unit)
• multiple inlets and outlets for chamber air
• resistance and mixing in extract silencers
• distributed intakes on chiller sides ends
• discrete, swirling outlets on chiller top
• Flow inside chiller not solved for

18
CFD Model Description - 2

• Dependent variables solved for
• pressure (total mass conservation)
• axial, lateral and vertical velocity components
• air/chiller effluent mixture temperature
• air residence time in chamber
• turbulence kinetic energy
• turbulence energy dissipation rate
• Independent Variables
• 3 spatial co-ordinates (x,y,z) and time

19
CFD Model Description - 3

• Iterative guess and correct solution procedure
  to convergence of scheme
• Typical domain size - 15x8x19 m.
• Around 1500 sweeps of domain required for
  convergence
• Typical nodalisation level - 207,000
• Convergence involves solution of around 2,500
  million simultaneous linked differential equations

20
CFD Model Description - 4

• The set of partial differential equations is
  solved within the defined solution domain and on
  a prescribed numerical grid
• The equations represent conservation of mass,
  energy and momentum
• The momentum equations are the familiar
  Navier-Stokes Equations which govern fluid flow

21
CFD Model Description - 5

• The equations may each be written in the form
• D(r j) /Dt div (r Uj - Gj gradj ) Sj
  
• Terms cover transience, convection, diffusion and
  sources respectively
• Equation is cast into finite volume form by
  integrating it over the volume of each cell

22
IAC Ltd Chiller Test Chamber Ventilation Study

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed
• Results Obtained
• Conclusions

23
Supply / Extract Arrangements Studied

• Chamber air supply arrangement
• Straight supply ducts
• Angling of supply end regions
• Blocking middle region
• Chamber air extract arrangement
• Long side outlet ducts
• Small additional centre outlet
• Large centre outlet
• Small vestigial side outlets

24
Chamber Geometry Arrangements Studied

• Effect on chiller intake temperatures of
• Friction on walls ceiling
• Silencer pressure losses at inlet outlet
• Mid-height wall lip
• End hood on chiller
• Baffles along chiller sides
• Lip around centre ceiling extract
• Swirl breaker above chiller

25
Chiller Operating Conditions Studied

• Chamber dimensions
• Dimensions 2.2 x 8..7 x 2.44 m. high
• 12 outlet fans, swirl angle 30 degrees
• Chiller Hot Operating Condition
• Inlet temperature 35 deg.C
• Heat input 951kW or 1019 kW
• Air flowrate 75 or 67 m3/s
• Chiller Cold Operating Condition
• Inlet temperature 7 deg.C
• Heat input -363kW

26
Chamber Operating Conditions Studied

• Chamber Air supply Rate
• Initially 110 of chiller throughput
• ( i.e. 1.175 82.5 m3/s
• Subsequently increased to 90 m3/s
• Hot Operating Condition
• Inlet supply temperature 35 deg.C
• Cold Operating Condition
• Inlet supply temperature 7 deg.C

27
Overview of Workscope

• 34 simulations performed in 7 stages
• Stage 1 - Original design concept effect of
  swirl add small central outlet remove lateral
  offset longer central outlet add wall friction
  hot cold runs
• Stage 2 - chamber outflow partitioning
  sensitivity effect of inclining and
  part-blocking some of supply inlets
• Stage 3 - revised chiller inflow partitioning
  Central outlet lip and vestigial side outlets

28
Overview of Workscope

• 34 simulations performed in 7 stages
• Stage 4 - chiller swirl level outlet silencer
  resistance central outlet lip.
• Stage 5 - Increase chamber air rate chiller end
  and side baffles increase chiller heat rate and
  reduce throughput for worst case.
• Stage 6 - Worst case run with swirl breaker
• Stage 7 - Air loading run with chiller off.

29
IAC Ltd Chiller Test Chamber Ventilation Study

• Industrial Context
• Objectives of Study
• Benefits of using CFD
• Description of CFD Model
• Simulations performed
• Results Obtained
• Conclusions

30
Original Design Concept

• Configuration
• Two long low-resistance side outlets
• No central outlet
• Supply
• Chamber supply rate 82.5 m3/s
• Chamber supply temp 35 deg. C
• Hot Chiller Operating Condition
• Chiller throughtput 75 m3/s
• Temperature rise through chiller 11.17 deg C

31
Original Design ConceptPredictions - 1

• Max temperature difference across chiller intake
  ports 7.99 oC
• Min intake temperature 35.1 oC
• Max intake temperature 43.1 oC
• Mean intake temperature 36.65 oC
• Mean intake residence time 7.46 sec
• Max chamber residence time 72.1 sec

32
Original Design Concept Predictions - 2
33
Original Design Concept Predictions - 3
34
Original Design Concept Predictions - 4
35
Original Design Concept Initial Findings

• Hot, highly swirling flow from chiller outlet
  creates non-symmetric flow patterns in chamber,
  despite symmetry of inlet, outlet and chiller
  locations
• Hot recirculating flow re-entrained into chiller
  end intakes, creating a hot end and a cold
  end
• Intake temperature differences are eight times
  desired criterion ...

36
Stage 1 Simulations

• Effect of chiller outlet swirl level
• reducing swirl improves matters
• (but this is not an option)
• add small central outlet
• DT reduced to 4.35 oC
• lengthen central outlet
• DT increases slightly to 4.85 oC

37
Stage 1 Simulations

• Chamber wall friction
• DT reduced slightly to 4.52 oC
• Hot and Cold Operation
• DT for cold operation about half that when hot
• Hot operating condition will thus be the worst
  case for achieving the chiller intake temperature
  uniformity criterion

38
Stage 2 SimulationsSupply/Extract geometry
sensitivity

• Chamber outflow partitioning sensitivity
• Tinkering with outlet resistance does not improve
  matters. DT in range 4.5 to 6oC
• Inclining the outer supply inlets
• Directing outer inlet jets towards chiller ends,
  to sweep away descending hot fluid from intakes,
  does hot have desired effect.
• DT in range 4.5 to 5oC

39
Stage 2 SimulationsSupply/Extract geometry
sensitivity

• Blocking the lower centre supply inlets
• Blocking the central lower inlet increases the
  incoming momentum of supply jets towards chiller
  sides . Does hot have very dramatic effect,
  reducing DT by about 0.1oC

40
Stage 3 SimulationsSensitivity to Chiller
Inflow specification

• Chiller inflow partitioning (ends, sides, base)
  derived from
• Manufacturers Estimates
• For hot operation, DT is about 5.3oC
• IAC Experimental Measurements
• For comparable run, DT is about 3.1oC
• these more reliable data used for subsequent
  simulations

41
Stage 3 SimulationsExtract geometry
sensitivity

• Central outlet lip
• Adding a deep lip around periphery of central
  roof outlet should allow capture of more of
  swirling flow from chiller top.
• Unfortunately, it also provides a shortcut for
  hot air to the ends, leading to a dramatic
  increase in DT !
• Moral Not all intuitive aids work as one might
  expect . . .

42
Unexpected Outcomes . . .
43
Unexpected Outcomes . . .
44
Stage 3 SimulationsExtract geometry sensitivity

• Further enlarged central outlet with vestigial
  side extract ducts
• Long extract ducts on each side replaced by four
  smaller apertures at intervals central outlet
  further enlarged, but no lip. DT falls to about
  2.7oC

45
Stage 4 Simulations Sensitivity to chiller
outlet swirl level

• For a reference geometry hot operation
• effect is dramatic . . . . .
• for 0 swirl, DT is about 0.6oC
• for 30 swirl, DT is about 1.3oC
• for 100 swirl, DT is about 6.0oC

46
Stage 4 SimulationsFurther i/o geometry
sensitivity

• Outlet silencer resistance
• Specification of high and low resistance zones in
  inner and outer regions of central outlet has
  small (10 reduction) effect on DT
• Increase airflow from 82.5 m3 to 90 m3
• Increasing ventilation rate has a greater effect,
  reducing DT by about 25

47
Stage 5 SimulationsSensitivity to internal
baffles

• Side and end baffles added, to
• channel supply air to intakes at chiller ends
• prevent descending hot air plume being
  re-entrained into end inlets
• Baffles and shrouds do not perform quite as
  envisaged . . .
• Dead zones form in end shrouds, negating some of
  supply-air channelling benefit .
• However, DT reduced by about two thirds

48
Baffled and Shrouded- - Tried and rejected - -

49
Baffled and Shrouded- - Tried and rejected - -
50
Baffled and Shrouded- - Tried and rejected - -
51
Baffled and Shrouded- - Tried and rejected - -
52
Final SimulationsWorst case Operating Scenario

• Chamber air flow increased to 90 m3/s
• Chiller air flow reduced to 67 m3/s
• Chiller heat input increased to 1051 kW

53
Final Design Concept

• Geometry
• No baffles or end hoods
• Normally-directed air supply
• Side-wall ridge at mid-height
• Enlarged central extract with four small extracts
  along each side
• Shallow centre outlet lip
• Swirl Breaker fitted between chiller top and
  air extract duct

54
Final Design Concept wall roof tiles
removed for clarity
55
Final Design Concept wall roof tiles
removed for clarity
56
Final Design Concept wall roof tiles
removed for clarity
57
Final Design ConceptPredictions - 1

• Max temperature difference across chiller intake
  ports 0.96 oC
• Min intake temperature 35.01 oC
• Max intake temperature 35.97 oC
• Mean intake temperature 35.13 oC
• Mean intake residence time 4.27 sec
• Max chamber residence time 128 sec

58
Final Design Concept Predictions - 2
59
Final Design Concept Predictions - 3
60
Final Design Concept Predictions - 4
61
Final Design Concept Predictions - 5
62
Final Design Concept Air flow predictions -
axial plane
63
Final Design Concept Air flow predictions -
transverse plane
64
Final Design Concept Air flow predictions -
transverse plane
65
Final Design Concept Air flow predictions -
transverse plane
66
Final Design Concept Air flow predictions -
plan view
67
Final Design Concept Air flow predictions -
plan view
68
Final Design Concept Air flow predictions -
plan view
69
Final Design Concept Residence time
Considerations

• Flow-averaged residence time of air in chamber is
  19 secs. Maximum predicted is 128 seconds.
• Region located below mid-wall lip, at non-control
  panel end and side, is the slowest clearing dead
  zone.
• Times contoured depict time following injection
  into domain

70
Final Design Concept Residence time predictions
71
Final Design Concept Residence time predictions
72
Final Design Concept Residence time predictions
ataxial section through centre of chiller
73
Final Design Concept Residence time predictions
at transverse section through centre of chiller
74
Final Design Concept Residence time predictions
at transverse section through far end of chiller
75
Conclusions - 1

• Attainment of 1-degree or less variance in
  chiller intake temperatures is thwarted by the
  re-entrainment of the hot swirling plume issuing
  from the top.
• Attempts to modify air flow patterns to rectify
  matters by tinkering with inlets, outlets,
  baffles etc. only met with partial success

76
Conclusions - 2

• Breakthrough came in controlling the influence of
  the highly swirling chiller -outlet flow, by use
  of a waffle-iron type of swirl-breaker device.
• This was more effective than using measures to
  try to divert the flow further downstream.

77
Conclusions - 3

• Final design concept can meet the clients design
  criterion for acceptable variance in chiller
  intake temperatures.
• Some fine-tuning may be required upon final
  installation

78
Closure

• Final Concept

• Initial Concept

79
P.S. . . . . . .

• And so they went ahead and built the test chamber
  . . . . .

80
The fanfare of trumpets
81
P.S. . . . . . .

• But the bean counters said
• lets try to do without the swirl breaker
• and verily the measured results fell short
• of the clients design specification .
• and so they put the swirl breaker back,
• and came back for more modelling ,
• to fine tune the design

82
The Test Chamber as built
83
Changes to model for as built test cell
geometry

• Smaller 10-fan unit, located symmetrically
• and reversed
• Chiller intakes uniform flux along sides and
  bases of units, but not at ends
• Anti-clockwise swirl at chiller fan outlets
• Prescribed, non-uniform fan outlet temperatures
  intake values computed
• Non-uniform fan swirl profile

84
Changes to model for as built test cell
geometry

• Domain extended upwards to include representation
  of outlet plenum
• Non-uniform inlet flow distribution, based on
  measured values
• Triangular-section wall protrusions
• Fine mesh swirl breaker

85
Model for as built test cell geometry
86
Changes to model for as built test cell
geometry

• Chamber air flow up to 92.5 m3/s
• Chiller throughflow up to 72m3/s
• Chiller heat input down to 869 kW
• Grid nodalisation up to 276,000 cells

87
As Built Test Cell Geometry Fine tuning
simulations

• Effect on intake temperature profile of
• Uniform and non-uniform Tfan distributions
• Swirl breaker fitment
• Fan swirl angle

88
Changes to model for as built test cell
geometry

• Run Specification
• Non-uniform fan outlet temperatures, based on
  experimental measurements. Maximum temperature
  54.4oC
• Minimum temperature 42.6oC
• Fan swirl angle 45o
• As built swirl breaker design

89
As built test cell geometryFlow pattern at
chiller intakes
90
As built test cell geometryTemperatures at
intake level
91
As built test cell geometryFlow patterns at
fan outlet level
92
As built test cell geometryTemperatures at
fan outlet level
93
As built test cell Flow patterns below
swirl-breaker level
94
As built cell Temperatures below
swirl-breaker level
95
As built test cell Flow patterns at central
outlet lip level
96
As built cell Temperatures at central
outlet lip level
97
As built cell Flow patterns at section
through fans 3 4
98
As built cell Temperatures at section
through fans 3 4
99
As built cell Flow patterns at section
through fans 7 8
100
As built cell Temperatures at section
through fans 7 8
101
As built cell Flow patterns at section
through near-side fans
102
As built cell Temperatures at section
through near-side fans
103
As built cell Flow patterns at section
through far-side fans
104
As built cell Temperatures at section
through far-side fans
105
As built test cell Chiller intake
temperature profile
106
As built test cell Chiller intake
temperature profile
107
Conclusions

• Highly non-uniform fan discharge temperatures
  lead to inlet temperature variations along length
  of unit
• Local fluctuations can exceed 1 degree,
  especially close to top of unit
• However, mixed-mean values for each unit remain
  well below this criterion

108
THANK YOU FOR YOUR ATTENTION
109
Points of Contactfor further information

• Flowsolve Ltd
• Dr. Paddy Phelps
• Dr. David Glynn
• 130 Arthur Rd.
• Wimbledon Park
• SW19 8AA
• 0208 944 0940
• cfd_at_flowsolve.com

• IAC Ltd
• Mr. Geoff Howse
• Mr. Greg Smith
• IAC House
• Moorside Road
• Winchester
• Hants SO23 7US
• 01962 873000