Data Centre Regulation

(1)

WPs 2.1, 2.2, 2.3, 2.4, 2.5

(2)

i-STUTE cooling based projects

WP2.1. and WP2.2 Supermarket

refrigeration

WP2.3 . Data centres

WP2.4. Transport refrigeration

WP2.5. Integrated heating and

cooling

Cost of ownership

Carbon/ energy

Materials, resources & waste

(3)

WP 2.1 and 2.2 Retail refrigeration

Background

• 40-70% of energy in supermarkets used for refrigeration

• UK retail refrigeration ~ 9-10 TWh/year

–

~75% chilled, ~25% frozen

• 1.5% of UK energy used by retail

• ~7.3 Mt CO2 (~26% direct, ~74% indirect)

• Temperature control, carbon

emissions increase at consumer

end of cold chain

Deliverables

• Refrigeration road map

• State of the art display cabinet

(4)

WP 2.1 Retail chilling and freezing

• WP2.1.1 – Technologies will be initially investigated

and sifted

• WP2.1.2 – In parallel with WP2.1 technologies will be

investigated with a proof of concept prototype

• WP2.1.3 – Non technical barriers preventing uptake,

will be assessed ie customer reaction,

implementation, cost-benefit, incentives

• WP2.1.4 –A trial of the prototype in-store with ASDA

WP

(5)

Technologies investigated

• 77 technologies evaluated at last meeting

• New technologies added:

– Ejectors

– Flooded evaporators

– 2-stage compression

– Turbine expansion machines

– Fan motor outside cabinet

– Lights outside cabinet

– Defrost drain traps

– Integral distributed system

– Thermostatic flow control

– Air deflectors/guides

– Improved axial fans

– Diagonal fans

– Defrosts (additional information)

– Dual port TEV/TXV

– Glazing (additional information)

– Efficient HE design

(6)

The model

• Supermarket model further

developed

• Store modelled - ASDA

Weston-Super-Mare

• Typical large supermarket

• Model can be adapted to

different store sizes and

configurations

• Further information obtained

from City Holdings (ASDA

contractors)

• However, stalled recently due to

(7)

(8)

• Emissions per year:

– Direct = 343.8 tCO

_2e

– Indirect = 373.7 tCO

_2e

• Ratio indirect : direct = 1.1

• High direct as very high refrigerant

charge (~1000 kg in cabinet

circuits)

• Leakage rate medium (~10% per

year)

• Therefore effect of changes to

refrigerant have high impact on

CO

_2e

emissions

(9)

DIRECT EMISSIONS

(10)

INDIRECT EMISSIONS

(11)

DIRECT (grey bubble) AND INDIRECT EMISSIONS (open bubble)

(12)

Next steps

• Need further data and clarification from ASDA:

–

Cooking appliances

–

Some costs for applying technologies

–

Some data still is not logical

• Based on current information possible to halve emissions with

paybacks of less than 3 years

• If technologies with less than 3 year paybacks were applied and

assuming application of:

–

Cheapest options with best paybacks

–

Minimum savings applied

–

Simplest option (where more than one option available)

• the carbon savings would be:

(13)

WP2.1 Deliverables

• Contact with CSEF, agreed to

create dynamic supermarket

model with team at Brunel

• Keynote for ICEF12 (Quebec)

• Opportunity to publish book

from road map work

• Peer reviewed paper on

technological options (IJR)

(14)

WP2.3 - Data Centre Cooling

14

Background

• Data centres currently account for approx. 2-3% of

total electricity consumption in the UK

• Typically, approx. 50% of data centre energy is used

for cooling and humidification

• Data centres are generally air cooled and the

heat dissipated to ambient

• Limited focus on heat recovery

Deliverables

• Roadmap/report on cooling

• Detailed investigation - integrated cooling, heat

recovery and heat transfer.

(15)

Roadmap

There are 4 main approaches to data centre

cooling:

Remote air cooling:

- Using CRACs or CRAHs/chilled water. Also air and water economisation

15

Local air cooling:

- Close coupled cooling e.g. rack rear door chilled water heat exchanger

Direct liquid on-chip cooling:

- Water or dielectric cold plate heat exchanger in direct contact with electronic components

Total immersion liquid cooling:

(16)

Roadmap

Comparison of 4

main cooling

approaches:

16 Cooling Method Characteristics

Remote air Local air Liquid direct-to-chip Liquid immersion Coolant(s) (P) Air (S) R/G/W/Chw (P) Air (S) Chw (P) 60/70% Liq (P) 30/40% Air (P)100% Liq Typical inlet-return temperatures Air: 25-35°C Chw:10-20°C Air: 25-35°C Chw:10-20°C Liq:40-60°C Air: 25-35°C Liq: 40-60°C Heat capacity of

primary coolant Low Low High High ∆T Chip to coolant High High Low Low Heat recovery possible? Air  Chw  Air  Chw  Air  Liq  Liq  Heat reuse value Low Low High High

(17)

Roadmap

Future trends:

1. Increasing data centre efficiency:

- Growth in data centre capacity, size and processing speed to meet user needs big data, internet of things (IoT).

- Increasing numbers of high performance computing (HPC) and hyperscale servers, and development of exascale data centres

2. Greater utilisation of IT server resources:

- Increased IT work levels, server consolidation and virtualisation, move to cloud - Software defined data centres (SDDC)

17

3. Chip and server architecture

development:

- Further miniaturisation of ICs – down to 1 nm scale by 2020s

- Adoption of 3D architecture e.g. vertical chip stacking with microchannel liquid cooling, especially for memory chips

Microliquid heat sinks between stacked dies

(18)

Data centres and District heating networks

• Currently supply only 2% of heat

demand in UK by district heating

• UK government plans to

substantially expand district

heating networks making use of

waste heat sources e.g. data

centres

18

• London plans to build a low temperature heat network – supply

temperature 70°C (London Mayor reports, 2012; 2013)

• Data centre waste heat could be upgraded via heat pumps to

contribute heat at this temperature

(19)

19

• Largest number of (colocation)

data centres is in London

(approx 75)

• Majority are concentrated in

central London, along the

Thames

(20)

London heat use and district heating networks

20 (Map from: http://tools.decc.gov.uk/nationalheat map/) (Map from: http://www.londonheatmap.org.uk/Mapping/)

• Yellow lines indicate existing

heat networks, red lines indicate

proposed heat networks

(21)

Detailed investigation of cooling and waste heat

recovery in data centres

Objectives:

• To construct a test facility to simulate a conventional IT server rack

(~5kW)

• To investigate a range of cooling methods, environmental conditions

and waste heat recovery systems

21

• To evaluate the quantity and quality of recovered waste heat, for

different cooling methods

• To investigate the carbon

and cost implications of

increasing waste heat

temperature to e.g. 70°C

using heat pumps

(22)

Details of IT server rack test facility

• 1 x 42U standard server rack ~2 m (h) x 0.6 m (w) x 1.07 m (d)

• 42 x 1U servers, total weight approx. 500 kg (10-15 kg per server)

• Linux operating system and benchmarking software to provide

adjustable, constant heat generation for all servers

• IT servers instrumented with thermocouples, velocity and humidity

sensors. Measurement of total energy input to servers and cooling

equipment and heat recovered

(23)

Next steps

Activities Duration Deliverables Due date

Finalise, format and publish roadmap report

May- June 2015 Roadmap report 1st_{July 2015}

Publish journal paper on waste heat recovery from data centres

May-June 2015 Journal paper 1st_{July 2015}

Construct data centre test facility and commission

May-Dec 2015 Operational facility Report on test facility construction 1st_{Dec 2015} 1st_{Feb 2016} Experimental testing of cooling and heat recovery methods

Dec 2015-end of project

First interim report Additional interim reports Final report 1st_{May 2016} TBC End of project 23

(24)

WP 2.3 Deliverables

• Internal report on cooling of data centres –

October 2014

• Initial internal heat recovery report –

December 2014

• Report/roadmap of Future technologies with

input from Robert Tozer – March 2015

• Dissemination – paper on data centre waste

heat recovery to be presented to CIBSE

technical symposium April 2015 at UCL

• Journal paper on heat recovery drafted

• Detailed heat recovery study commences

January 2016

(25)

Background

• UK primary food distribution by RRT uses 40% more

energy than non-refrigerated vehicles

• Environmental Impact

• Indirect emissions -

• Transportation - 2 Mtonnes of indirect CO₂ emissions from the engine alone.

• Refrigeration - ????

• Direct emissions -

• RRT units leak up to 30% of their total refrigerant charge per year

• System Durability & Reliability

Deliverables

• Development of a model to investigate direct and

indirect emissions

• Optimising system performance

(26)

Research Plan

1. Investigate different types RRT vehicle

technologies

2. Analyse maintenance and leakage records

to:

a) Identify problematic components/ sources of refrigerant leakage

b) Suggest generic solutions for leak tight systems

3. Develop a model to;

a) Estimate direct/ indirect carbon emissions b) Evaluate the effectiveness of various

concepts

4. Measure actual RRT data

5. Validate and optimise model

(27)

27

Project Plan flow chart

PhD Thesis Conduct Prelim

Study & Data Analysis I Validate & Optimize Model Report for Transport Industry Collect Data & Analyse Develop Model

(28)

Model Development – Refrigeration Performance

28

2. A preliminary model to predict the performance of RRT systems has been developed.

• MS Excel Mathematical model

• Focuses on typical last-mile RRT vehicle (i.e. small vans to medium rigid

refrigerated trucks) used for urban distribution.

• Calculates relative proportion of various refrigeration heat loads and corresponding indirect carbon

emissions:

i. Wall transmission

ii. Natural infiltration due to gaps, cracks iii. Door infiltration

iv. Product load

(29)

29

(30)

Challenges and Solutions

30

Issues with direct drive RRT units:

• Large amount of heat entering during

door openings

• Refrigeration system stop working when vehicle stops

=> system is off when load at its highest

• Running time between stops may be

short

=> time insufficient for temp pull-down

Common solutions include:

• Oversize unit; use door protection; employ a hybrid system

Planned approach: Determine optimum design

• For the average load profile

(31)

31

Project Schedule

31

Today

Develop Model - May 2014 - Jan 2016

Data Collection

-• Meeting with Fleet Owner (Data Supplier) – May 2015 • Initiate data collection –Jun 2015

(32)

WP 2.4 Deliverables

• _{Internal report on leakage - Feb 2014, August 2014} • _{LSBU Registration document -RES2 – June 2014} • _{Summer school conference June 2014, June 2015} • _{Internal report on modelling platforms- August 2014}

• LSBU Literature review internal report-Res 3B – Oct 2014

• Internal report on modelling platforms- August 2014, Nov 2014

• Ethics application approval -Jan 2015

• LSBU Annual Report RES 4 – April 2015

• Impact Hubbard have changed their system design to minimize leakage.

(33)

Background

• To investigate the interactions of underground railway tunnels and ground heat exchangers • To investigate the potential indirect use of waste heat from the tunnels to heat buildings

above ground.

Deliverables

• Development of a model • Case study materials

(34)

2. Project time line with the key milestones

Stage 1 & 2 Stage 3 Stage 4 Stage 6 Stage 7

(35)

Figure 1 Annual temperature

distribution within soil Figure 3 Numerically simulated tunnel and ground surface temperatures

Figure 2 BHE wall temperatures versus tunnel proximity

(36)

WP 2.5 Deliverables

• _{Internal reports April July and November 2014, February 2015} • _{Summer school conference, Poster – June 2014}

• _{Registration document RES2 – September 2014} • _{Literature review internal report – October 2014}

• _{Internal report on modelling platforms - November 2014} • _{Registration document RES3 – February 2015}

• Conference paper submission – February 2015

• Manuscript submission to a Journal – March 2015

(37)