Max Protect – Max Efficiency
Engineers Design Guide to Large UPS
C. Mayo Tabb Jr.
Senior, 3-phase Regional Manager
June 2014
DCUG Spring 2014 Survey Results
What are your top three (3) facility / network concerns?
0.0%
10.0%
20.0%
30.0%
40.0%
Availability (uptime)
33.6%
50.0%
38.8%
Adequate monitoring / data center…
Protect
50.9%
50.5%
32.1%
Heat density (cooling)
Energy efficiency (energy costs &…
42.1%
38.8%
22.4%
Power density
19.4%
Space constraints / growth
48.6%
Efficiency
28.0%
27%
15.8%
18.7%
Security (physical or virtual)
25.5%
22.4%
Technology changes / change management
21.2%
24.3%
Data center consolidations
16.4%
12.1%
Data storage
9.7%
9.3%
Regulatory compliance
Other
60.0%
1.8%
Customers want both:
Efficiency without
Compromising
Availability
6.5%
Spring 2014
Spring 2013
Capacity & Efficiency
Capacity &
Efficiency
driving data
center change
Source: Uptime Institute / 2012 Symposium
Max Protect-Max Efficiency
 Max Protect
–
–
–
–
–
–
Availability of power to load is top priority
Data is unique and cannot be recovered
Initial cost and operating cost are secondary
Configuration and batteries are equally important
Tier3 & 4 where every chance of failure must be eliminated
Typically wet cells or 20 year VRLA battery
 Max Efficiency
– Initial cost and operating cost are top priority
– Data can be recovered or process repeated
– Availability are secondary
– Site and configuration redundancy
• Designed to tolerate a failure
– Typically 5/10 year VRLA battery
Emerson Network Power
Max Protection and Max Efficiency UPS
June
2014
500, 625, 750, 800, 900, 1100kVA
Liebert NXL
250, 300, 400kVA
SMS, 1+N, N+1
Eco-Mode, Intelligent Paralleling
Maximum Protection
400, 500, 600kVA
Liebert NX
225, 250, 300kVA
SMS, 1+N
Eco-Mode, Capacity on Demand (Softscale)
March
2014
Maximum Efficiency
Liebert eXL
625, 750, 800 kVA
1200 kVA
SMS
Eco-Mode, Capacity on Demand (Softscale)
3phase In / 3phase Out
200
800
Capacity, kVA
1200
1600 kVA
System Availability
UPS Design Engineer’s Quote
“An isolation transformer hides many rectifier
and inverter sins”
“A transformer increases cost, footprint and
lowers efficiency”
What saves
What fails
AC
Output
MBB
FBO
BIB
EG
BFB
A
E
FBO
CB2
N
N
MBJ
EG
Trap Disconnect
GEC
isolated
MIB
3P
AC
Non isolated other
than 480V
Keep the load up
To Batteries
Isolation
Transformer and Transformer-Free UPS
Liebert Products
Characteristic
AIC
Paralleling
Transformer-Free
Max Efficiency
NX - eXL
Transformer-Based
Max Protect
NXL
65k,100k
65k,100k or 150k
1+N, (1+N &N+1)
1+N, N+1
PDU Start/bolted short
Input / DC / Output Isolation
Voltages
Alarm on Acid leak
480v,HRG
480v,600v,HRG Opt.
95-97% / 98%
92-94% / 98%
UL1778 4th Edition,
OSHPD,FCC
UL 1778 4th Edition,
OSHPD,FCC
Weight / Size
Efficiency – double
conversion/eco-mode
Agency Listing
Liebert NXL
Enterprise-Scale UPS Protection for Medium/Large Data Centers
250kVA/225kW 480/575/600VAC
300kVA/270kW 480/575/600VAC
400kVA/360kW 480/575/600VAC
500kVA/450kW 480VAC
625kVA/625kW 480/575/600VAC
750kVA/675kW 480/575/600VAC
800kVA/800kW 480VAC
1100kVA/1100kW 480/575/600VAC
1100kVA/1100kW
 Greater than 1,348 units under warranty
and service contract, 24,683,136 Hrs.
 MTBF = 6,170,784 Hrs.*
 Best field MTBF of any Liebert UPS
1125kVA/1125kW
 4 times improvement over Legacy UPS
Units in Blue provide DC isolation
Maximum Protection UPS System
AC
Output
MBB
FBO
BIB
EG
12P isolated
A
BFB
3P
CB1
AC
FBO
E
Trap
Disconnect
CB2
N
N
MBJ
EG
or
GEC
12P non isolated
To Batteries
NXL800 Rectifier
Liebert NXL
 Ratings to 1100kVA/kW
 Transformer-based
– 600v without add-on transformers on
DC isolation versions
 Efficiency
– 94+% Dual Conversion
– 98+% Active Inverter
Intelligent Ecomode
– System level Intelligent Paralleling
Transformer
-Based
Monolithic
Constructio
n
MIB
Liebert NXL
Industry Leading Performance
Superior
Stack up
Performance
UL STD. 1778
4TH Edition
Handles Faults
High, Flat Efficiency Curve
User Friendly DSP Controls
Leading Power Factor Capability
Liebert NXL High Efficiency Modes of Operation,
“Intelligent EcoMode”
 Intelligent EcoMode
– Increases efficiency
by running the bypass
in parallel with the
inverter.
– If poor quality AC
detected, switches to
full dual conversion
mode
• Outage
Bypass
AC Input
Double Conversion Operation
Static Switch
Inverter
Rectifier
Rectifier
AC Input
Battery
Intelligent EcoMode
• Transfer
Bypass
AC Input
Static Switch
Inverter
Rectifier
Rectifier
AC Input
Battery
•
•
•
•
•
Bypass source is monitored
Inverter in on
Inverter matches bypass
Load harmonics profiled
Efficiency gain
NXL,NX,eXL Configurations
System Level Static Switch and Controls
SS
I
R
Single Module System (SMS)
NXL,NX,eXL
SS
R
I
BB
SS
R
I
BB
Distributed Bypass (1+N)
NXL,Nx,eXL
Cost Effective Design
Product Line Scope
BB
I
BB
R
I
BB
R
I
BB
R
I
BB
Centralized Static Switch (N+1)
NXL,eXL
Highest MTBF Design
R
SS
1+N (Distributed Static Switch)
N+1 (Central Static Switch)
NXL and eXL share N+1 SCCC
Slightly less costly
ASCO
 3200-5000 amp Continuous-duty Static
Slightly more reliable
Switch SCCC
 1000% Overload rating
 De-rates at 1600,2000,2500,3000
amps but costly
 N+1 UL-1558 & UL-891 to 200 kaic
 1+N UL-15558 &UL-891 to 100 kaic
13
Liebert NX,
Transformer-free UPS System
 225kVA/225kW 480V
(Fixed Capacity or SoftScale to 300 kVA/kW)
 250kVA/250kW 480V
(Fixed Capacity or SoftScale to 300 kVA/kW)
 300kVA/300kW 480V
(Fixed Capacity ) Best price point
 400kVA/400kW 480V
(Fixed Capacity or SoftScale to 600 kVA/kW)
 500kVA/500kW 480V
(Fixed Capacity or SoftScale to 600 kVA/kW)
 600kVA/600kW 480V
(Fixed Capacity ) Best price point
8500 units installed
in Europe since 2007
14
Liebert NX,
Large Transformer-free System
 Transformer-free, 480 Volt, 3-wire design
 Unity PF rating, kW = kVA
 Leading/Lagging PF load support
 Configurations:




Single-module systems
Parallel 1+N systems, to 6 Modules
Dual bus systems
Common Battery option for 2 modules
 95% efficient in dual conversion
 98-99% efficient in eco-mode

High overload capability
(125 %10 min, 150% 1 min)
 High power density / small footprint
 UL 1778 Edition 4 listed
 Liebert Service coverage/capability
 Life.net automatic “call home” monitoring
 Field mtbf 1.2M hours,8500 installed since
2007 by European methodology
 OSPHD tested
Liebert NX600
Dual or single input; optional input CB
Input Jumpers
For single input
16
100 kAIC Withstand Rating
 Fuses provide a 100 kAIC withstand rating.
 3 wire +G input/output only – no 4 wire
Input
fuses
 No output breaker or option for one
 Unit will always be with external MBC
 MIB or MOB/IOB provides disconnect
NX600
starting
800kVA
PDU w SS
pulse
NX600 3phase
bolted fault w/o
bypass –unit kept
running after
breaker opened
NX600 Technical Data
S610 450/500
NXL 450/500
NX600
798 amps input
804 amps input
761 amps input
Combined effect of efficiency and
advanced PWM rectifier optimized to
VRLA Batteries
25% battery recharge obsolete
10x recharge rate obsolete – 20X
VRLA batteries life is shortened if
fast recharge – 5% is typical max
Max Efficiency Liebert eXL UPS!
Pushing Double Conversion Efficiency to 97%
Leading power factor loads without de-rating 0.7 leading to 0.7 lagging
2 level vs 3 Level
NPC2
2 Level
NPC1
Inverter Topology Comparison 400VAC
2L, NPC1, and NPC2
2L, Legacy, NX, Powerware 9395, MGE G7K – 94-95%
IGBT Losses
NPC1, APL, APM, Mitsubishi <250kVA, -95-96%
NPC2, eXL, Mitsubishi >250kVA, GE – 96-97%
Switching Frequency
UPS System Efficiencies
100.0%
Efficiency
95.0%
Liebert eXL
Active Inverter
Intelligent
EcoMode*
90.0%
Liebert eXL
Dual Conversion**
85.0%
80.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
Load
*Current Estimate **Subject to upward revision
Liebert eXL
Input section
AC Input
DC input
Draw out logic
and customer Fuse protected
options
100kaic SCCR
23
EXL800
Dual 400kW cores
Draw-out for ease
of service
Boast Charger
Phases A,B,C
8 IGBT packs per phase/core
Core
inductors
24
eXL
Output and Static switch
Static switch SCR
Output
Bypass input
Output and BFB
breakers
25
EXL800
Cooling design for maximum efficiency
 High Efficiency is increasingly effected by fan losses
– Fan kW are a larger portion of total losses at higher efficiency
 4 x 600 cfm ball bearing 50,000 hour fans per core
 Fan failure is alarmed via tack signal from fan
 Shutdown/bypass determined by temperature
 100% load – 35 degrees C at 800 kW
– Continuous operation requires all fans
 Up to 90% load- 35 degrees C at 800 kW
– Continuous operation with one failed fan
 Above 90% load
 Operates until temperature bypass/shutdown on failed fan
Remember the battery
It is responsible for half the load losses!
Expected Yearly Replacement
40%
35%
Percentage
30%
25%
20%
15%
10%
5%
0%
1
2
3
4
5
6
Years
7
8
9
10
Battery Life Comparison
eXL800
Technology
Minutes
Warranty
Life
Cost
4x HR540
VRLA
6
3 yr
3-5 yr
$87k
4x HX5500
VRLA
5
3 yr
3-5 yr
$87k
4x XE-95
Pure lead
3
2 yr
2-4 yr
$102k
5x XE-95
Pure lead
6
5 yr
5-6 yr
$127k
5x HR7500
VRLA
17
4 yr
4-6 yr
$150k
4x HX925
VRLA
16
3 yr
3-6 yr
$156k
AVR95-33
VRLA - stack
6
5+15
12-15 yr
$176k
DXC-23 san
wet 1.250
15
3+17
12-18 yr
$178k
DXC-27 poly
Wet 1.215
15
3+17
12-18 yr
$237k
AVR4100
VRLA - stack
27
5+15
12-15 yr
$253k
5x Li-on
LiFeMgPO4
6
10 yr
10+ yr
$311k
* Replacement cost at 75% in year 4-5, 8-10 etc.
*
Battery
*
Alber – individual cell monitoring
The Difference – Early Detection of Failures
 Typically, internal resistance
increases slowly over time
and use
 Early detection allows for cell
replacement to avoid load
loss
Resistance Trend
 AC impedance testing will
detect a bad cell
 Only when very close in time
to when the cell is failing or
has failed
Impedance Trend
40%
35%
30%
Percentage
Why is it 4 years
for a 10 year VRLA?
Expected Yearly Replacement at 77 F
25%
20%
15%
10%
5%
0%
 10 year design life in telecom float test
1
2
3
4
5
6
7
8
9
10
Years
– 24 cells16 amps for 8 hours versus 240 cells 450 amps for 5 minutes
– Warranty – 3 years full+7 years pro-rata
 Year 4 – 2%x240 cells=5 cells
–
–
–
–
One fails every 2-3 months for 1 string
For two strings one every month
For 4 strings one every two weeks - 5 cells between 90 day PM’s
IT will barely tolerate this number of service calls
 Year 5 – 15%x240 cells=36 cells
– One fails every 10 days for 1 string
– For two strings one every 5 days
– For 4 strings one fails every 2 days – 36 cells between 90 day PM’s
– IT will Not tolerate this number of service calls
How many cell failures before replacement ?
Liebert Battery Mean Time
Between Failure (MTBF) Study
Outages Per Million Hours
0.45
Battery Maintenance
(No Monitoring)
Experience: High reliability
0.40
0.35
Alber On-site
0.30
Experience: Significantly longer
runtime before a failure
0.25
0.20
Ntegrated Monitoring
0.15
Experience: No outages due to
bad batteries
0.10
0.05
0
0.00
On-Site PMs
On-Site PMs
with Alber
Ntegrated
Monitoring
Integrating remote and on-site service:
 Remote Service – Centralized Technicians
 On-Site Service – Field Technician
 Monthly PM’s (1 on site / 11 Alber Monitoring)
2010* Study based on batteries under
Liebert contract from battery strings with a
total
of 9.5 million run hours prior to the end of
their expected service life.
*Updated 2013