Fire risk in MTBF evaluation for UPS system

RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE Fire Risk in MTBF Evaluation for UPS System Stefano ELIA, Alessio SANTANTONIO Department DIAEE – Electrical Engineering Section, Sapienza University, Piazzale Aldo Moro 5, 00185 Rome, Italy [email protected], [email protected] DOI: 10.15598/aeee.v14i2.1662 Abstract. The reliability improvement of no-break redundant electrical systems is the first aim of the proposed strategy. The failure of some UPS (Uninterruptible Power Supply) system may lead to the fire occurrence. The most used electrical configurations are presented and discussed in the paper. The innovation of the proposed method consists of taking into account the fire risk to improve the accuracy of wiring configuration and components’ failure rate. Thorough research on MTBF (Mean Time Between Failure) data has been performed for each wiring component and UPS. The fire risk is taken into account introducing an equivalent fire block in the Reliability Block Diagram scheme; it has an MTBF value calculated form yearly statistics of UPS fire events. The reliability of the most used UPS electrical configurations is evaluated by means of the RBD method. Different electrical systems have been investigated and compared based on MTBF. The importance of fire compartmentation between two or more UPS’ connected in parallel is proved here. Keywords Electrical installation, failure rate, fire risk, MTBF, no-break power system, RBD, redundant electrical system, reliability, UPS. 1. Introduction Ten years of maintenance activity in no-break electrical installations reveals a lot of design errors and a lack of reliability. Currently, it is common to evaluate the reliability only for the equipment and for electrical system’ components. Today, it can be noticed that there is a defect in reliability evaluation of all project choices. Many real fire case occurrences on UPS were caused by power electronic or battery fail. The producers do not consider fire hazard significant for the UPS device and perform tests only in standard and good maintenance conditions. The UPS machines are very vulnerable and subjected to inadequate maintenance, overvoltages, high temperature, working conditions, and other electrical system malfunctions. Nowadays redundant UPS’ are not isolated by a fire compartment. No-break systems are often designed following a wrong guideline. The standard procedure considered for reliability improvement consists of a simple installation of two UPS’, which are connected in parallel. UPS power electronic, Control Unit, manual and static bypass, batteries and other system components are usually installed in the same room. All devices are, therefore, exposed to the same fire risk. Working UPS unit may be involved in fire event caused by the failure of another UPS unit in the same room, thus rendering the system redundancy ineffective. Moreover, an emergency manual bypass wired out of the UPS room is never installed. In those conditions a fire event entails certainly the complete failure of the no-break system. A short and contained fire is also sufficient to generate smoke and risk of toxic air; in this case nobody can access the UPS room and technicians are obliged to communicate immediately with the director to inform him of the imminent failure and recommend to stop all current operations. A lot of fire case studies have been investigated; the complete failure of the no break electrical system was often due to the lack of fire compartmentations between two UPS. 2. Reliability Model The reliability evaluation for each electrical configuration is based on the Reliability Block Diagram model (RBD) [1], [2]. By means of this method, the MTBF of UPS, Control Unit, batteries, switches, and other components are represented. Data on UPS fire occurrence frequency were obtained during ten years of consultant activity in the hospitals. In about a hundred of case studies the existence of two UPS’ room fire events is c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 189 RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE proved per year (2 % per year). Two hypotheses are necessary to evaluate the failure rate [3] by statistical data on fire risk. Firstly, the failure rate is constant in time. Secondly, break components are not repairable but quickly replaceable (MT T R = 0). The average fire failure rate is defined as the ratio between number of fires and number of studied events per time. It is shown in Eq. (1): λAV G =NF NT OT ·T,(1) where NFis the quantity of fire events, NT OT is the number of observed systems and Tis the observation time. Subsequently the MT BF value is calculated [4] in Eq. (2): MT BFF=1 λAV G =NT OT ·T NF =100 ·8760 2= = 438000 h. (2) Using the Eq. (2), the reliability calculation can be based on different fire statistics. Moreover, calculation can be developed implementing a parametric analysis varying the value of fire statistic. The reliability evaluation model is used based on these hypotheses. Firstly, UPS is only considered as a no break system that avoids voltage dips. Secondly, the continuous energy source is based only on power supplier’s grid or emergency diesel generator. 3. MTBF Data MTBF data on studied components have been deduced by an accurate statistical survey. Used data were obtained from: Gold Book [5], some papers [1], [6], [7], [8], [9] and many datasheets. The UPS’ MTBF is evaluated including the presence of the onboard automatic static bypass and batteries. MTBF value for the Fire Risk Factor block is pointed out by the maintenance activity experience, Eq. (2). The average values of MTBF for all components are pointed out in Tab. 1. Tab. 1: Average MTBF of no-break system’s components. Components Symbol Failure rate [failure·h−1]MTBF [h] Battery (lead acid) BAT 8.52086·10−71173590 Circuit breaker CB 4.348·10−6229991 Complete UPS module (internal STS and batteries included) UPS 1.3779·10−572574 Control Unit CU 1.33333·10−6750000 Fire Risk Factor FRF 2.28311·10−6438000 Inverter INV 0.00002 50000 Rectifier REC 0.00002 50000 Static Transfer Switch STS 9.79499·10−6102093 Switchgear Bus Bar SBB 1.08334·10−6923068 4. MTBF Evaluation of Various No-Break Electrical Systems 4.1. One UPS The reliability of a base configuration with only one UPS is shown in this subsection. UPS ROOM UPS SHORT-BREAK BUSBAR CB CB NO-BREAK BUSBAR LOW VOLTAGE SWITCHGEAR Fig. 1: No-break system with one UPS. The Reliability Block Diagram of the one UPS configuration is shown in Fig. 2. CB FRF UPS SBB CB SBB Fig. 2: RBD scheme of no-break system with one UPS. The calculation shown in Eq. (3) reveals the total MTBF for the system configuration with only one UPS. MT BFT= 1 1 MT BFSBB +1 MT BFCB +1 MT BFF RF +· · · 1 · · · 1 MT BFU P S +1 MT BFCB +1 MT BFSBB = = 37140 h= 4.2years. (3) 4.2. Two UPS Without Fire Compartmentations The configuration with two UPS‘ connected in parallel is considered here. Each UPS has a rated power greater than the load demand. Machines are installed in the same room together with the batteries and without any fire compartmentations. The respective configuration scheme is shown in Fig. 3. The RBD scheme of that c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 190 RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE UPS1 UPS ROOM UPS2 SHORT-BREAK BUSBAR CB CB CB CB NO-BREAK BUSBAR CU LOW VOLTAGE SWITCHGEAR Fig. 3: No-break system with two UPS without fire compartmentations. system is shown in Fig. 4, where it must be highlighted that the Fire Risk Factor of both UPS‘ influences the entire system. CU CB CB UPS2 SBB CB CB UPS1 SBBFRF1 FRF2 Fig. 4: RBD scheme of no-break system with two UPS without fire compartmentations. The MTBF for that configurations is can be computed with the use of Eq. (4), Eq. (5), Eq. (6) and Eq. (7): MT BFU P S = =1 1 MT BFCB +1 MT BFU P S +1 MT BFCB = = 44494 h. (4) RUP S1U P S2=RUP S1+RU P S2−RU P S1·RU P S2,(5) where R[8] is the reliability and it is defined as R=e − 1 MT BF ·t. MT BFU P S1UP S2=Z∞ 0 RUP S1U P S2·dt = = 66741 h. (6) MT BFT= =1 1 MT BFSBB +1 MT BFF RF 1 +1 MT BFF RF 2 +· · · 1 · · · 1 MT BFCU +1 MT BFU P S1UP S2 +1 MT BFSBB = = 43385 h= 5 years. (7) Only one Control Unit is usually installed for an emergency load switching between two UPS‘, that choice makes the MTBF worse. 4.3. Two UPS in Different Fire Compartments The reliability of two fire compartmented UPS‘ is studied here. In Fig. 5 and Fig. 6, there are shown the system configuration and the RBD scheme respectively. UPS1 UPS ROOM1 UPS2 UPS ROOM2 SHORT-BREAK BUSBAR CB CB CB CB NO-BREAK BUSBAR CU LOW VOLTAGE SWITCHGEAR Fig. 5: No-break system composed of two UPS with fire compartmentations. CU CB CB FRF2 UPS2 SBB CB CB FRF1 UPS1 SBB Fig. 6: RBD scheme of no-break system with two fire compartmented UPS. In this case, each Fire Risk Factor is related exclusively to the respective UPS. Calculations to evaluate c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 191 RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE the total MTBF are represented in Eq. (8), Eq. (9), Eq. (10) and Eq. (11): MT BFU P S =1 1 MT BFCB +1 MT BFF RF +· · · 1 · · · 1 MT BFU P S +1 MT BFCB = 40391 h. (8) RUP S1U P S2=RUP S1+RU P S2−RU P S1·RU P S2.(9) MT BFU P S1UP S2=Z∞ 0 R12 ·dt = 60587 h.(10) MT BFT=1 1 MT BFSBB +1 MT BFCU +· · · 1 · · · 1 MT BFU P S1UP S2 +1 MT BFSBB = = 49987 h = 5.7 years. (11) 4.4. Two UPS‘ and One STS with Fire Compartmentations Another system improvement consists of installing a safety external bypass over two UPS‘. All these devices must be installed in different fire compartmented rooms. In the Fig. 7 and Fig. 8 the configuration and the RBD scheme are shown respectively. According to the proposed system scheme, calculations to evaluate total MTBF are shown in Eq. (12), Eq. (13), Eq. (14), Eq. (15), Eq. (16), Eq. (17), Eq. (18) and Eq. (19). MT BFU P S =1 1 MT BFCB +1 MT BFF RF +· · · 1 · · · 1 MT BFU P S +1 MT BFCB = 40391 h. (12) RUP S1U P S2=RUP S1+RU P S2−RU P S1·RU P S2.(13) MT BFU P S1UP S2=Z∞ 0 R12 ·dt = 60587 h.(14) MT BFU P S1UP S2CU = =1 1 MT BFU P S1UP S2 +1 MT BFCU = 56058 h.(15) UPS1 UPS ROOM1 UPS2 UPS ROOM2 STS STS ROOM CB SHORT-BREAK BUSBAR CB CB CB CB CB NO-BREAK BUSBAR CU LOW VOLTAGE SWITCHGEAR Fig. 7: No-break system made up of two UPS and one STS with fire compartmentations. CU CB CB FRF2 UPS2 SBB CB CB FRF1 UPS1 SBB CB CB STS Fig. 8: RBD scheme for no-break system with fire compartmented UPS and STS. MT BFST S = =1 1 MT BFCB +1 MT BFST S +1 MT BFCB = = 54080 h. (16) RUP S1U P S2CU ST S =RUP S1U P S2CU + +RST S −RUP S1U P S1CU ·RST S .(17) MT BFU P S1UP S2CUST S = =Z∞ 0 RUP S1U P S2CU ST S ·dt = 82616 h.(18) c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 192 RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE MT BFT= =1 1 MT BFSBB +1 MT BFU P S1UP S2CUST S +· · · 1 · · · 1 MT BFSBB = 70070 h = 8 years. (19) The total MTBF for compartmented system is 8 years. Calculations have been also performed for the case of absence of fire compartmentations. Installing two UPS‘ and one STS in the same room results in MTBF of 6.4 years. 4.5. Three UPS‘ with Fire Compartmentations The system configuration made of three UPS‘ installed in different rooms is studied here. The configuration scheme is shown in Fig. 9. UPS1 UPS ROOM1 UPS2 UPS ROOM2 UPS3 UPS ROOM3 CB SHORT-BREAK BUSBAR CB CB CB CB CB NO-BREAK BUSBAR CU LOW VOLTAGE SWITCHGEAR Fig. 9: No-break system made of three fire compartmented UPS. The respective RBD scheme is shown in Fig. 10. The total MTBF of the three UPS‘ configuration is computed using the Eq. (20), Eq. (21), Eq. (22), CU CB CB CB FRF2 UPS2 SBB CB CB CBFRF3 UPS3 FRF1 UPS1 SBB Fig. 10: RBD scheme for fire compartmented three UPS configuration. Eq. (23), and Eq. (24). MT BFU P S = =1 1 MT BFCB +1 MT BFF RF +· · · 1 · · · 1 MT BFU P S +1 MT BFCB = 40391 h. (20) RUP S1U P S2=RUP S1+RU P S2−RU P S1·RU P S2.(21) RUP S1U P S2UP S3=RU P S1U P S2+RU P S3− −RUP S1U P S2·RUP S3.(22) MT BFU P S1UP S2U P S3= =Z∞ 0 RUP S1U P S2UP S3·dt = 76743 h.(23) MT BFT=1 1 MT BFSBB +1 MT BFCU +· · · 1 · · · 1 MT BFU P S1UP S2U P S3 +1 MT BFSBB = = 60494 h = 6.9 years. (24) The total MTBF for three compartmented UPS‘ is 6.9 years. In case of absence of fire compartments for the same configuration the reliability is studied; and MTBF of 5.1 years is obtained for three UPS installed in the same room. With respect to the configuration with two UPS‘ and one STS, a little decrease of reliability is to be noticed. This is due to the better MTBF of the STS compared to the UPS. On the opposite, this three system UPS permits all maintenance operations during working activities. 5. Conclusions Revised statistical data on MTBF components used in no-break systems have been summarized here. A c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 193 RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE method to take into account the fire risk in a Reliability Block Diagram model has been performed. The reliability of seven different UPS configurations has been studied by means of the RBD method. The total MTBF has been computed for each configuration taking into account the effect of the devices’ fire compartmentation. The results of this comparative analysis are shown in Fig. 11. Fig. 11: MTBF for different no-break power systems by fire compartmentation. These MTBF results can be also converted to yearly failure probability values by means of Eq. (25): F P%=1 MT BF ·100.(25) These results demonstrate the importance of fire compartmentation for reliability improving in redundant UPS systems, especially for hospitals and safety systems. The more suitable configuration consists of two UPS and one STS fire compartmented, which achieves the best MTBF. Moreover, the three compartmented UPS‘ configuration and permits a complete maintenance during system’s operation. Acknowledgment In this part of the article the authors can express their gratitude to projects from the results of which the article has been written, or to people who have contributed to the results published in the article. It is not allowed to insert any figures or logos here. References [1] RAHMAT, M. K., S. JOVANOVIC and K. L. LO. Uninterruptible Power Supply (UPS) system configurations: Reliability comparison. In: 2010 IEEE International Conference on Power and Energy. Kuala Lumpur: IEEE, 2010, pp. 835–840. ISBN 978-1-4244-8947-3. DOI: 10.1109/PECON.2010.5697695 [2] CHIESA S. Affidabilita, sicurezza e manutenzione del progetto dei sistemi. 1st ed. Turin: CLUT, 2008. ISBN 978-8879922647. [3] Military Handbook: Electronic Reliability Design Handbook, MIL-HDBK-338B. Washington: Department of Defense, 1998. [4] NIL, M., M. NIL and B. CAKIR. MTBF analysis in OEM Company: Applications to ZCZVT PWM soft-transition inverters. In: 9th International Conference on Environment and Electrical Engineering. Prague: IEEE, 2010, pp. 353–356. ISBN 978-1-4244-5370-2. DOI: 10.1109/EEEIC.2010.5489940. [5] IEEE Std 493-1990. Recommended practice for the design of reliable industrial and commercial power systems. New York: IEEE, 2007. ISBN 978-07381-7083-1. [6] RAHMAT, M. K., S. JOVANOVIC and K. L. LO. Reliability Comparison of Uninterruptible Power Supply (UPS) System Configurations. In: Telecommunications Energy Conference ’Smart Power and Efficiency’ (INTELEC). Hamburg: IEEE, 2013, pp. 1–6, ISBN 978-3-8007-3500-6. [7] RAHMAT, M. K., S. JOVANOVIC and K. L. LO. Reliability and Availability modelling of Uninterruptible Power Supply (UPS) systems using Monte-Carlo simulation. In: 2011 5th International Power Engineering and Optimization Conference. Shah Alam: IEEE, 2011, pp. 267–272. ISBN 978-1-4577-0355-3. DOI: 10.1109/PEOCO.2011.5970403. [8] BODI, F. “DC-grade” reliability for UPS in telecommunications data centers. In: INTELEC 07 - 29th International Telecommunications Energy Conference. Rome: IEEE, 2007, pp. 595–602. ISBN 978-1-4244-1627-1. DOI: 10.1109/INTLEC.2007.4448849. [9] RAHMAT, M. K., S. JOVANOVIC and K. L. LO. Reliability Modelling of Uninterruptible Power Supply Using Probability Tree Method. In: Proceedings of the 41st International Universities Power Engineering Conference. Newcastle: IEEE, 2006, pp. 603–607. ISBN 978-186135-3429. DOI: 10.1109/UPEC.2006.367549. c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 194 RELIABILITY AND SAFETY OF ELECTRONICS DEVICES VOLUME: 14 |NUMBER: 2 |2016 |JUNE About Authors Stefano ELIA was born in Rome, Italy. He received his Degree in 1999 and the Ph.D. in 2003 in Electrical Engineering from Sapienza University of Rome. Now is employed as a Professor at the Electrical Engineering Institute of Sapienza University. His research interests include power electronics, electrical systems, maintenance, reliability, safety, measures, energy saving and renewable energy. Alessio SANTANTONIO was born in Caserta, Italy. He is an Electrical Engineer graduated at Sapienza University in 2014. Now is employed as laboratory assistant at the Electrical Engineering Institute of Sapienza University. He collaborates with some electricians’ companies on maintenance procedure for electrical power systems. His research interests include power systems, power quality, reliability, maintenance. c 2016 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 195