Fluids Dynamics Simulations - Ansys UK/staticassets/SM... · Fluids Dynamics Simulations Stephen...

Fluids Dynamics Fluids Dynamics SimulationsSimulationsFluids Dynamics Fluids Dynamics SimulationsSimulations

Stephen McCormickStephen McCormick

Technical ServicesTechnical Services

ANSYS UKANSYS UK

Stephen McCormickStephen McCormick

Technical ServicesTechnical Services

ANSYS UKANSYS UK

© 2011 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary© 2011 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary

Content

• A simple question

• Guest presenter: Karl Else, Hurley Palmer

Flatt

• Guest presenter: Dan Jestico & Ben Abel,

Hilson Moran Partnership

© 2011 ANSYS, Inc. All rights reserved. 2 ANSYS, Inc. Proprietary

Hilson Moran Partnership

• Some examples

• Solar loads – simple or not

• External flows – the next step in turbulence

• Conclusions

A Simple Question

ANSYS CFD offers cutting edge tools...


But do you need it?

Airflow is predictable?

• Complex and non-intuitive

• Depends on the interaction of multiple

features

Which situation will see the

© 2011 ANSYS, Inc. All rights reserved. 4 ANSYS, Inc. Proprietary1m/s 1m/s


highest velocity?

A B

And the answer is….


highest velocity?

A B


1m/s 1m/s

A B

• Karl Else


• Dan Jestico & Ben Abel

Some Examples...Some Examples...Some Examples...Some Examples...


Internal Ventilation

• Occupant comfort

– Natural, forced,

mixed mode

• Air freshness


• Humidity control

• Odour control

• Equipment

performance

External Flow

• Pedestrian comfort

• Pressure mapping

– Structural loading

– Natural ventilation


– Natural ventilation

• Intake/ outlet siting

• Exhaust dispersion

Courtesy: Peter Brett Associates

Data Centres

The Cooling Approach

But,

too often

cabinet

cabinet

cabinet

21oC

45% RH

+10-20oC

Hot

aisle

Cold

aisle

Cold

aisle

fan

CRAC


• CRAC units supply cold air to a floor plenum

• Cold air distributed to cold aisle inlets to computer

racks/ cabinets

• Racks exhaust heated air to hot aisles

14oC Floor Plenum

Two rack layouts:

different performance


Maximum Intake Temperature per Rack

What is happening in the plenum?


Plenum Airflow Patterns

Tile Airflow

Distribution

Static Pressure Distribution

Ducting Ducting OptimisationOptimisationDucting Ducting OptimisationOptimisation

Courtesy of ArupCourtesy of ArupCourtesy of ArupCourtesy of Arup


Nature of the problem

Stadium bowl cooling


65000 seat semi-outdoor stadium

Air delivery system design

Middle-East location

• Equalisation of the flow rates

at the seat supply nozzles

– 36 types of plenums => 36

individual solutions

– each serving 300-1000 slots

• High cost for each evaluation

The challenge


• High cost for each evaluation

– Mesh size ~ 3 - 6 million cells

– Run time ~ 2 - 6 hrs on 8

CPUs

Solution approach

• Pressure equalisation using diffusing

plates

• Geometry and the position of the plates

have big influence on performance

– Computational optimisation study is

required

– Definition of plate geometry, size,

location creates 2-6 parameter


location creates 2-6 parameter

optimisation problem

Optimisation strategy

• Goal of optimisation is to minimise

standard deviation of nozzle flow rates

• Use of statistical models to approximate

response surface

– Search using standard methods

(genetic algorithms)


(genetic algorithms)

– Statistical model guides selection of

new evaluation points (consolidate or

explore)

– Vastly reduces number of evaluations

ResultsResults


• Benefit to Arup’s Client: Significant cost reduction

as pressure balancing grilles at the slots (65,000)

are not required anymore

• No other feasible way than optimization

algorithms to find the optimum plate

configurations

Summary


configurations

• Run time: 36 different plenum types each

requiring 30-50 CFD models

– Each CFD model requires

• 1-2 hours meshing time

• 2-6 hours CFD simulation time

Solar Load Modelling


Solar Loads:

The Challenge

• Potentially dominant heat source

• Strong spatial influence


• Strong spatial influence

• Radiation models are expensive

Solar Loads:

The Solution

• Solar Load Model

– ‘Simple’ ray-trace

approach

– Local heat flux


– Calculates in seconds

Glazed Facades

Glass Properties - IGDB

Window v Glazing System Thermal and Optical Properties

ID : 45

.

.

Layer Data for Glazing System '45 CFD SHGC Single Clear'

ID Name D( ") Tsol 1 Rsol 2 Tvis 1 Rvis 2 Tir 1 Emis 2 Keff

------ --------------- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ----

Outside

103 CLEAR_6.DAT #0.225 .771 .070 .070 .884 .080 .080 .000 .840 .840 .578

Inside


Inside

.

.

Angle 0 10 20 30 40 50 60 70 80 90 Hemis

Vtc : 0.884 0.883 0.882 0.879 0.872 0.852 0.804 0.688 0.427 0.000 0.805

Rf : 0.080 0.080 0.081 0.083 0.089 0.106 0.152 0.267 0.528 1.000 0.144

Rb : 0.080 0.080 0.081 0.083 0.089 0.106 0.152 0.267 0.528 1.000 0.144

Tsol : 0.771 0.770 0.767 0.761 0.750 0.727 0.680 0.575 0.346 0.000 0.689

Rf : 0.070 0.070 0.070 0.072 0.077 0.093 0.134 0.239 0.484 1.000 0.128

Rb : 0.070 0.070 0.070 0.072 0.077 0.093 0.134 0.239 0.484 1.000 0.128

Abs1 : 0.159 0.160 0.163 0.167 0.173 0.180 0.185 0.186 0.170 0.000 0.173

SHGCc : 0.818 0.817 0.815 0.811 0.801 0.781 0.735 0.630 0.396 0.000 0.740

.

.

Glass Properties

IR: εεεε=0.84?


Resulting Temperatures (oC)


Flow Velocity (m/s)


Dealing with Solar Gains

• ANSYS CFD Solar Load Model: practical solution

– Uses standard manufacturer’s data

– Spatial heat gains calculated in seconds

• Can be coupled with full radiation models


• Can be coupled with full radiation models

– E.g. detailed glazed unit analysis

– Full physics response

– Need to be confident in supplied properties

External Atmospheric Flows


Courtesy: Peter Brett Associates

Atmospheric Flows are Turbulent


Smallstructures

Largestructures

Turbulence Models

Zero Equation

RANS (inc Unsteady)In

cre

asin

g A

ccura

cy, In

cre

asin

g C

ost


RSM

LES

DNS

Incre

asin

g A

ccura

cy, In

cre

asin

g C

ost

Embedded LES:

Making LES Practical

ELES Region Box


Full Model Size: 1000m x 1000m x 200m

ELES Box Size: 104m x 140m x 64m

Outer Domain:

Non-Conformal Interface

LES region

requires finer

mesh resolution

than the outer

RANS region


RANS region

Flow Domain decomposed into

two regions with either

conformal or non-conformal

mesh interface in between

Inner Embedded LES Box Region

Boundaries are located sufficiently far away from the building of

interest so that:

- the synthetic turbulence can be developed

- wake from the upstream building is resolved using LES


Upstream Building

Main Building

Resolved Flow Velocity


Contours of Velocity Magnitude (m/s)

Resolved Pressure and Velocity


Key Facts:

• Embedded LES (ELES) can be used for urban built

environment

• Typical time to purge flow through LES region – 30s

• ~1 week to compute (4s per day)


Conclusions


Conclusions

Conclusions

• CFD is a proven technology

– Influences

– Verifies

– Explains


– Explains

• New challenges keep coming

– Capabilities keep evolving

• ANSYS CFD offers cutting edge tools

And the answer is….

Adverse

pressure

gradient


Velocity Pressure

Fluids Dynamics Simulations - Ansys UK/staticassets/SM... · Fluids Dynamics Simulations Stephen...

Documents

Transcript of Fluids Dynamics Simulations - Ansys UK/staticassets/SM... · Fluids Dynamics Simulations Stephen...