October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

27
October 2006 Andrea Gaddi, Physics Dept CMS Infrastructures Issues CMS Infrastructures Issues

Transcript of October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

Page 1: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

CMS Infrastructures IssuesCMS Infrastructures Issues

Page 2: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Basic Detector InfrastructuresBasic Detector Infrastructures

Powering

Cooling

Cabling

Safety issues

Access facilities

Definition of <Detector Infrastructures> is wide,ranging from very basic services to more detector specific

systems.We give here some highlights on design guidelines

and main parameters of CMS Infrastructures,focusing in particular on:

Page 3: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Detector PoweringDetector Powering

Different power utilities:

Power to Front End Electronics (FEE)Power to Counting Rooms and Site Control CentresPower to auxiliaries & services

Different power sources:

Uninterruptible Power (battery back-up)Secured Power (diesel-generator back-up)Non-secured Power

Page 4: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Power distribution at P5Power distribution at P5

QAC

Page 5: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

How much powerHow much power

System Installed power (kVA) Rated power (kW)

General services on site 5,250 2,200Electronics racks 4,000 2,300LV to FEE 2,000 1,000Magnet + Cryogenics 3,300 800 (1,250)*Ventilation stations (inc. smoke extraction) 5,850 1,250 (3,000)*Surface cooling stations 6,500 4,000Underground cooling stations: 2,000(water) 600(C6F14) 600 (900)*

Total 28,900 12,750 (15,250)*

(*) refers to transient operations (cooling down, powering up, etc.)

Page 6: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Low Voltage PowerLow Voltage Power

Schematics of LV distribution

6x300 kVAisolationtransformers

AC/DCconverter(ECAL)

QuickTime™ et un décompresseurPhoto - JPEG sont requis pour visualiser

cette image.

RSS

UPS

PM54

UL55

2MVA18 kV -400 V

SUX5

SE5

18 KV

UXC55USC55

2 MVA - 2 minutes

400V DC

230V AC

230V AC

Courtesy S. Lusin

Page 7: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Power LossesPower Losses

Courtesy S. Lusin

Consider a factor 2 wrt final end user to design transformers and power lines

Page 8: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Detector CoolingDetector Cooling

Front End Electronics CoolingElectronics Racks CoolingCaverns Ventilation

Cryogenics

Page 9: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

FEE Cooling at CMSFEE Cooling at CMS

More than 1 MWatt is dissipated into heat by CMS Front Electronics boards.This large amount of heat needs to be transferred away from the Detectorvia appropriate cooling fluids (water or CxFy, depending on working temperature).

CMS has 6 independents cooling loops, serving the following systems:

Muon Endcaps water 16 deg 100 kWthMuon Barrel water 16 deg 50 kWthHCal + Yoke Barrel water 16 deg 60 kWthECal water 16 deg 300 kWthRacks system water 16 deg 1600 kWthTracker, Pixels, ES C6F14 -15/-30 deg 150 kWth

Page 10: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Cooling InfrastructuresCooling InfrastructuresChilled water at 6 deg is produced on surface and dispatched to thedifferent cooling stations present on site (above and below ground)that finally produce water at 16 deg for the different cooling loops.This arrangement has made possible to test a significant part of CMSon surface, before lowering the Detector down into the cavern withouthaving s large impact on infrastructure costs.

Page 11: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Racks CoolingRacks Cooling

Turbine unit

Heat exchanger(s)

Air-deflector

Water manifold

Electronics crate

Axial fans

Vertical air-flow

Up to 8 kW per rackIntegrated smoke detectionIntegrated fire-quenching

Page 12: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

rack mounted PCs

air-to-water heat-exchanger

fans

Horizontal air-flow

PC Racks CoolingPC Racks Cooling

Up to 10 kW per rack (40 PCs)

Page 13: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Caverns VentilationCaverns VentilationHumidity and temperature controls: keep dew point as lower as possible

Page 14: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

CryogenicsCryogenics

The cryogenic plant at CMS site has the function to cool down andkeep at 4,7 K the 230 tons of the CMS Superconducting Coil.

The refrigerator system can deliver a cooling power of 800 W at 4,7 K,plus 4500 W at 60 K to cool the Coil thermal screens and in addition to that4 g/sec of L-He to cool the 20 kA Coil Current Leads.

Cooling the Coil down from ambient temperature takes 3 weeks,with a maximum thermal gradient inside the cold mass of 15 deg.

In case of quench, the temperature rises up to 70 K and 3 days arenecessary to bring the cold mass down to 4,7 K

A 6,000 lt L-He storage tank sits close to the cold mass to allow a slow-dischargefrom full current without warming up the coil.

Page 15: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Coil cooling downCoil cooling down

300

K

25 Feb 20063 Feb 2006

4 K

Page 16: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Detector CablingDetector Cabling

Cabling an LHC detector is one of the most time consuming and challenging tasks.As cables come rather late in the construction process,little or no contingency is given to perform this task.

CMS, due to its particular modularity, can be cabled by slices,making the process more flexible and repetitive.

We have some 20,000 cables on the detector itself, with an average length of 15 meters.In addition, there are some 4,000 cables that connect the detector to the adjacent

Counting rooms (average length 90 meters).

Think big!

Full-high false floor (1,80 m) below counting rooms and large cable-chains,make the cabling work a bit less painful than usual.

Page 17: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

On-Detector CablingOn-Detector Cabling

Courtesy G. Faber

Cable trays over the Coil cryostat on YB0

Page 18: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Off-Detector CablingOff-Detector Cabling

YE+1 and YB+2 chains under cabling

Page 19: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Cabling costsCabling costs

A few installation activities have their cost related to how your work is well organized as cabling.Manage cabling teams in terms of number of people on site, working on shift, requires perspective view.CMS has chosen to have teams coming from Institutions (Russia and China) namely foron-detector cables and place industrial contract with local cabling companies for off-detector cables.

On this basis we have reached good results, keeping costs at a reasonable level.

The predicted cost for Institution cabling was 2,3 CHF/meter and the actual one is 2,6 CHF/meterThe cost for cabling made by local companies ranges from 3 to 3,3 CHF/meter.Total estimated cost is about 1,7 MCHF (equivalent to 300 men/month).

Cabling cost is charged to sub-detectors according to the actual length of cables pulled.

Page 20: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Cabling organizationCabling organization

Page 21: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

SafetySafety

Major danger for any experiment is fire

o Fire prevention (rules & standards, gas inertion)o Fire detection (sniffers & gas analyzers, smoke-detectors)o Fire fighting (low pressure water, high pressure water, foam)

The power density of LHC experiments is much larger than any other before:2,6 GJ stored inside the Magnet

1 MW of FEE Low Voltage Power10 kW per rack

(…)CMS has been designed since the beginning with a great attention to prevent fireBeing the result of a fire inside the experimental cavern potentially disastrous, we

have tried to limit the overall risk keeping this possibility as low as possible.

Page 22: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Safety guidelinesSafety guidelines

Fire prevention:-- IS23 & IS41 Standards (Materials, Fire resistance, Smoke release)-- permanent N2 inertion inside vacuum tank and inner detectors Tracker, ECAL, HCAL-- all heat sources including front end electronics are intercepted by water,-- all detectors and their services (cables trays) must be neutral thermally with respect to their neighbors.

Fire detection:-- smoke detection in counting rooms and cable labyrinths (sniffers)-- smoke detection in detectors cuts electrical power to concerned racks and power supplies through DSS and its flexible action matrix-- smoke detection in racks with electrical cuts through DSS and local fire suppression for valuable electronics

Fire fighting:-- N2 flushing of muon chamber gaps in case of smoke detection-- passive foam system in case of loss control of a fire in the cavern-- low pressure water sprinklers in cable trays-- high pressure water sprayer to quench fire in PC racks and control rooms

Page 23: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Foam testFoam test

The UXC55 cavern has been filledwith hundreds of cubic meters offoam, blown out of the 8 nozzleslocated at cavern ceiling.

Page 24: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Detector Safety SystemDetector Safety System

The Detector Safety System is a PLC (Siemens) based supervisor that gets alarmsfrom its own sensors placed in every sensitive location. It has built insidean alarm/action matrix, that can be adapted to the evolution of the hardwareto be protected and/or the level of risk.Actions are taken with a time delay ranging from a few seconds to some minutes,depending on the severity of the risk.All its connections are hardwired and it has a built-in UPS to assure power anytime.It exchanges data with the detector slow control and has a back end interfacethat allows the user to visualize the status of all the sensitive parts.

Page 25: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Rack Safety SystemRack Safety System

Control touch screen to power racks on/off

Page 26: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

Access FacilitiesAccess Facilities

ScaffoldingFor long term planned works.They assure easy and safe access at high.

Personnel Access PlatformsThe more versatile access mean.Battery powered, can work insurface and underground areas.CMS has 6 PAC for installationand maintenance porpoises.

Page 27: October 2006 Andrea Gaddi, Physics Dept. CMS Infrastructures Issues.

October 2006 Andrea Gaddi, Physics Dept.

ConclusionsConclusions

Detector Infrastructures shall profit at a maximum of whatalready exists on the market. We have found on industrial

systems solutions to most of our needs.Our goal was to adapt them to

the complex needs of a high energy physics detector,rather than start from scratch with new solutions, that could

seem more convenient, but they often lack in reliability.

In short terms, do no reinvent the wheel!