Sie sind auf Seite 1von 11

2

Continuity of Supply Case Study


Krish Gomatom and Tom Short

C2.1 LEVELS OF CONTINUITY OF SUPPLY IN EUROPE [16]


The accuracy of the levels of continuity of supply depends on the nature of measurement practices. According to the 2005 benchmarking report published by CEER (Council of European Energy-regulators), measurement practices in European countries, vary widely. Due to the variations, the following should be noted: Though the scope of benchmarking interruptions has been extended to include short interruptions as well as long interruptions, not all countries separate their interruptions data into these two categories. Supply interruptions are measured in different ways. Continuity data may or may not be collected at all voltage levels. Continuity indicators may refer to all customers or be split between customers at different voltage levels. The most important factor to consider is that continuity indicators are not always defined in a comparable way. Continuity indicators can be weighted by three different methods: by customer, transformer or contracted power. This can give rise to differences depending on which weighting method is used. Based on the data compiled and report by CEER, interruptions shorter than three minutes are (separately) measured in only a few countries (Finland, France, Hungary, Great Britain and Italy). A group of countries is planning to measure short interruptions, for instance Lithuania and Czech Republic. Recently Norway has had companies record short interruptions, albeit at a less detailed level than for long interruptions. Since 2006 companies in Norway have been

Handbook of Power Quality Edited by Angelo Baggini 2008 John Wiley & Sons, Ltd

G
4 required to record and report short interruptions in the same scheme as long interruptions. This could be significant because of the increasing use of computers and other electronic equipment which is making customers increasingly concerned about short interruptions. CEER reported that, large differences were found across countries. Estonia, Ireland, Latvia and Slovenia which collect data at country level (note that in Ireland and Latvia there is only one distribution company). In all other countries continuity is monitored at a more detailed level: by distribution company, in Austria, Belgium (Wallonia), Czech Republic, Great Britain, Hungary and Norway; by administrative region in France and by concession in Sweden. A good number of the countries use a classification of territorial areas in order to distinguish between urban and rural (and thus set differentiated standards) (Table C2.1). Such distinctions are meant to capture technical differences among networks (overhead lines vs. underground cable, and so on). The criteria used for territorial classification vary from one country to another: Per number of inhabitants, at municipality level (Italy and Lithuania) or at locality level (France). Per number of customers, at municipality level (Spain and Latvia) or at locality level (Portugal). Other criteria: Greece (based on the distance from the nearest service center), Finland (based on the percentage of cable-5 categories), Ireland (by network configuration), Sweden (on the basis of meters of line per customer). In Italy the minimum time for notice is 24 hours in advance, otherwise the interruptions must be considered as unplanned. Should the planned interruption start before the notified timetable (more than 5 min in advance), it must be considered as unplanned. Should the
Table C2.1 Continuity indicators for distribution reliability (unplanned interruptions) among European nations

SAIDI, SAIFI and MAIFI per voltage level (H, M, L) SAIDI and SAIFI per voltage level (H, M, L) SAIDI and SAIFI per voltage level (H, M) SAIDI and SAIFI all voltages Average duration (D) and frequency (F) per contracted power or other

Other/no indicators

GB, HU, IT, NO CZ, GR, PT, FR, LT, NO SI (some data only), BE_Wallonia SE, EE, IE (SAIFI since 2006) AT (average D and F weighted on MV power affected, MV/MV, MV/LV) ES (average D and F weighted on MV power affected: TIEPI, NIEPI) FI (average D and F weighted on yearly energy consumption of the distribution area that one distribution transformer feeds) PT (TIEPI, ENS, excluding LV) NO (ENS, excluding LV: 1 kV) LV (number of interruptions), PL (no indicators)

G
5 planned interruption last longer than notified, the extra duration must be considered as unplanned. In Portugal the Commercial Relations Code, published by ERSE, establishes rules about the notice to the customer according to the reasons of interruption: Interruptions for reasons of public interest the entity responsible for the network must inform, whenever possible, and with a minimum prior notice of 36 hours, the customers which may be affected by the interruption. Interruptions for service reasons the entity responsible for the network has the duty to minimize the impact of the interruptions among customers. For this purpose, distributors may agree with the clients that will be affected the best moment for the interruption. If agreement is not possible, the interruptions must occur, preferentially, on Sundays, between 05:00 and 15:00, with a maximum duration of eight hours per interruption and five Sundays per year, per customer affected. The entity responsible for the network must inform with a minimum prior notice of 36 hours. Interruptions due to customer responsibility the supply interruption may only take place following eight days notice.

C2.2 HISTORICAL RELIABILITY TRENDS IN EUROPE [16] C2.2.1 Unplanned SAIDI


The duration of interruptions during the year is a key indicator of the quality of service that electricity customers receive. The length of an interruption can be influenced by a variety of factors such as the level of interconnectivity on the network, the voltage at which the interruption occurred, the distance to the fault, the accessibility of the fault, etc. Most countries calculate this measure based on users. According to the 2005 CEER benchmarking report there was a downward trend in the duration of interruptions for most European countries during 19992004. However, for a small number of countries the data is too volatile to identify any clear trends. One potential reason for the significant volatility is the impact of exceptional events on performance.

C2.2.2 Planned SAIDI


Planned duration relates to those minutes off supply experienced by customers where they have been given prior forewarning that they would be going off supply. Countries have their own rules as to how much notice is required to give customers in advance of planned work, with some countries counting intended planned work not notified in the correct timeframe as being unplanned work. Planned work is undertaken for a variety of reasons, such as to make improvements to the network, to connect new customers to the network, to make permanent repairs to the network, to carry out tree-trimming activities, etc. The 2005 CEER report shows a number of countries recording major increases in the average duration off supply due to planned work during 19992004. In the case of Hungary this is due to new constructions in order to replace old network elements to improve the supply quality.

G
6

C2.2.3 Unplanned SAIFI


The number of interruptions is a key indicator of the robustness of electricity networks and it is generally the case that the higher the voltage, the more robust the network. However, when interruptions do occur at higher voltages they tend to impact greater numbers of customers. Network operators have over time been seeking to reduce the number of interruptions experienced by customers and examples of ways to achieve this are: making the network more resilient, building in spare capacity known as redundancy and investing in network automation, which can result in customer interruptions moving from being classified as long interruptions (SAIFI) to short or momentary interruptions (MAIFI). The 2005 CEER data shows that, in most countries there appears to be a slight downward trend in the average number of interruptions during 19992004, which is understandable because once countries reach a certain level, it is generally the case that further improvements in performance are likely to be incremental. A number of countries have made significant strides in reducing the number of interruptions, with performance in Hungary and Italy dropping by over one interruption per customer in six years. Even some of those countries previously recording quite low levels of interruptions have been able to further reduce interruptions. Great Britain, the Netherlands and Spain have all seen significant percentage improvements in the number of interruptions. The impact of exceptional events is usually less pronounced for the number of interruptions compared to the impact on duration, although a number of specific atypical events in Italy in 2003 had a marked impact on the number as well as the duration of interruptions.

C2.2.4 Planned SAIFI


The number of interruptions due to planned work during 19992004, was somewhat volatile across the 13 countries providing data to CEER. The two countries experiencing major increases in the duration of planned interruptions also recorded increases in the average number of planned interruptions. In some countries the amount of planned work carried out on the networks has a less significant impact on the planned duration and interruption figures, as many network operators now employ techniques which enable them to work on networks whilst they are still live, therefore avoiding interruptions and minutes lost.

C2.2.5 Unplanned MAIFI


Some countries make no distinction between long and short interruptions and others do not collect any information on interruptions lasting less than 3 min. Additionally, not all countries differentiate between interruptions lasting less than 1 s, known as transient interruptions, and those lasting longer than 1 s and less than 3 min. The number of short interruptions can in certain instances give network operators advanced warning of developing problems which they can then address before they turn into long interruptions. As network operators invest in more automation and remote control it is likely that the number of short interruptions will increase. It may be that in the future customers demand limits on the

G
7 number of repeat short interruptions and a number of countries feel it is useful to keep track of this information at this time. Hungarian data for 2003 and 2004 complied by CEER, shows the total number of short interruptions per customer excluding exceptional events. Hungary splits short interruptions into transients (less than 1 s) and all other short interruptions (1 s to less than 3 min) but has found the data to be unreliable and has begun a project to ensure consistent short interruptions data from 2006.

C2.3 ECONOMIC ASSESSMENT OF RELIABILITY ALTERNATIVES C2.3.1 Textile Facility at a Rural Site
This case study is a small textile facility that produces custom reinforced textile products. The site is rural and has poor power quality. The main characteristics of the site are: Facility peak load = 500 kVA Utility service characteristics: The 34.5 kV distribution feeder has a total exposure of 80 miles (96 km) including 25 miles (40 km) of three-phase exposure. The facility is 15 miles (24 km) from the substation. Tree coverage is minimal. The substation has one 20 MVA, 8 % transformer with one other circuit. The nearest available feeder for a second feed is 10 miles (16 km) away. The lightning flash density, Ng = 2 flashes/km2 /year (which is moderate). The facility normally runs 24 hours per day, five days per week. Any disruption to the process line, which is made of cord covered with vulcanized rubber, will ruin the whole batch as only full-length runs are sellable. Two process lines treat the cord with the vulcanized rubber: the first, Treater 1, is impacted 40 times per year, and Treater 2 is impacted 30 times per year. Each treater process failure costs the facility $2500. The other significant equipment making up the process are an oxidizer and a high-frequency oven. The two treater lines are the most sensitive; short-duration interruptions to the oxidizer or oven will not scrap a batch. The summary of the loadings on each of the elements is: Treater 1, 80 kVA the main sensitive device in this line is an adjustable-speed drive. A PLC also controls some of the operations. Treater 2, 40 kVA the main sensitive equipment is a motor-starting contactor and two programmable linear controllers (PLCs). High-frequency oven, 100 kVA. Oxidizer, 80 kVA.

G
8 C2.3.1.1 Quality Levels The facility contains diverse sets of equipment; some are more sensitive than others. Table C2.2 shows estimates of the number of disturbances occurring at each quality level. This is a very poor utility location; the line is rural, high voltage and very long. A few months of monitoring have shown that the estimates are typical. C2.3.1.2 Comparison of Alternate Quality/Reliability Arrangements Several different categories of solutions are possible: Equipment-level solutions Facility process-level solutions Whole-facility solutions Utility-side solutions.

Equipment-level solutions can provide significant improvements. The adjustable-speed drive trips on an undervoltage relay. A drive add-on is available that adds voltage sag ride-through for sags down to 50 % by using the remaining voltage to keep the DC link voltage charged to appropriate levels. UPSs or constant voltage transformers will provide extra ride-through for the PLCs, and coil hold-in devices will improve the ride-through of the motor-starting contactor. These are not perfect improvements; they help with moderate voltage sags but not severe events. UPS or other voltage-conditioning equipment can support an entire process line. We consider two options: a flywheel-based UPS and a dynamic sag corrector. Utility-side solutions are unattractive in this case. A second utility line is too far away and the load is too small even to come close to justifying it. The facility is also too small for existing DVR equipment. By the same line of thought, facility-wide solutions are not particularly attractive. The sensitive load is well segregated and is easy to target. The cost and performance of each case are shown in Table C2.3. The options with the lowest total cost are the flywheel UPS and the equipment solution option. The UPS provides more complete protection and the lowest annual cost. The equipment solution provides the lowest initial cost. Almost any of the options are better than the do-nothing case. The UPS provides the most economic performance in this case because of all of the momentary interruptions. The equipment-level solutions, the dynamic sag corrector and the DVR do not protect against interruptions, and the feeder has many momentary interruptions.
Table C2.2 Estimates of the annual number of disturbances for different quality levels at the textile facility

Estimate Quality Quality Quality Quality Level Level Level Level 1 2 3 4 moderate sags severe sags/momentaries momentary interruptions long-duration interruptions 61 30 14 3

G
9
Table C2.3 Costs of various options for the textile facility

Configuration 1 No enhancements 2 Alternate feed with a fast switch 3 DVR 4 Equipment solution (ASD DC bus add-on, coil hold-in devices, UPS for PLCs) 5 Process solution with dynamic sag corrector 6 Process solution with flywheel UPS

Failure MTBF Annual downtime Rate (months) cost ($) 35 5 15 15 0 34 2 40 0 80 0 80 175 000 25 000 75 000 75 000

Annual equipment cost ($) 351 000 37 500 7 627

Total annual cost ($) 175 000 376 000 112 500 82 627

15

0 80

75 000

23 000

98 000

3 00

20 000

45 000

65 000

In situations where most of the disruptions are caused by voltage sags, the other options are much more attractive. C2.3.1.3 How Do Results Change with Facility Size? Both the equipment-level solutions and the process-level solutions scale well with size. The utility-side solutions start to become more economic at larger sizes, but still have trouble because of the existing poor performance and rural location. C2.3.1.4 Distributed Generation Options Distributed generation is not really an option at this facility because it does not have access to natural gas. Fuel must be shipped in, which makes continuous on-site generation much more expensive. If the facility were located in a spot with access to natural gas, distributed generation would be a very good option: the site already needs backup generators, and the large number of utility disturbances requires some in-plant ride-through improvement system.

C2.3.2 Hospital
This case study looks at a large hospital campus in a city in the southeastern USA. The total load is 2 MW. The medical campus has several buildings, including hospital care, lab testing facilities, billing and data processing, and a magnetic resonance imaging (MRI) center. Some of the important electrical characteristics are:

G
10 On-site facility generation is handled centrally. The buildings are wired well to support emergency power loads. Sensitive loads comprise 20 % of the facility load. Backup is applied locally with small UPS systems and is haphazardly done. The sensitive load consists of a variety of medical diagnostic equipment, the largest of which is an MRI scanner. A patient care information system has several terminals throughout the facility that access a central server. When this system is down, productivity drops measurably. The facility is in a suburban location. Two 12.47 kV feeders supply the hospital in a primary selective arrangement, each circuit coming from a different bus at the substation that is 3 miles (5 km) away. The substation transformers on both buses are 30 MVA units. The exposure on each circuit is roughly the same: 15 miles (24 km) total and 8 miles (13 km) three phase. The ground lightning density is 2 flashes/km2 /year. C2.3.2.1 Quality Levels Both of the feeders supplying the hospital are relatively good. Table C2.4 shows the estimated number of events at each quality level based on the prediction methods in Section 2.3 and the data in Section 2.4. C2.3.2.2 Comparison of Alternate Quality/Reliability Arrangements The existing protection against long-duration interruptions is excellent. Two utility feeds and backup generation provide an MTBF of approximately 200 years. The hospital is increasingly using information technologies, and the protection against short-duration interruptions is not nearly as good. The generators and redundant utility supplies provide no help for sags or momentary interruptions. The local UPSs are haphazardly applied: the UPSs are not regularly tested, some are not attached to emergency power outlets, and some sensitive equipment does not have UPSs. This leads to a relatively high unavailability on the UPSs. A good utility-side solution is a fast mechanical switch. An alternate feeder is already available, which greatly improves the economics. Using the fast utility-side transfer switch allows the hospital to eliminate many of the equipment-level UPS systems (the MRI scanner and the main server for the patient care information system still have UPSs in this situation).

Table C2.4 Estimates of the annual number of disturbances for different quality levels at the hospital campus

Estimate Quality Quality Quality Quality Level Level Level Level 1 2 3 4 moderate sags severe sags/momentaries momentary interruptions long-duration interruptions 16 8 3 1

G
11 Accurate outage or downtime cost is extremely difficult to quantify for hospital complexes, so it inappropriate to seek an optimal solution based only on minimizing the sum of the outage cost and equipment cost. Still, the cost and performance data from the SQRA analysis in Table C2.5 helps us choose an appropriate solution for the facility. We can use any of the following approaches alone, or in combination, to pick an appropriate solution: Performance target. If the facility identifies a performance target, the best solution is the lowest-cost solution that meets the target. If the lowest-cost solution exceeds budgetary capabilities, the performance target must be reduced. Budget limits. Facility budget restrictions limit the solution options. Cost and performance improvement. Any solution that improves performance at a lower equivalent cost than the existing system is a candidate for implementation. Cost/benefit ratio. Solutions with a better performance ratio (outages reduced versus cost) have higher priority. For the hospital, the choices are fairly limiting. Configuration 5, the base case plus a fast switch minus some equipment UPS, is clearly the best of these options. The fast switch centralizes some of the power quality improvement and reduces the need for as
Table C2.5 Performance and costs for several configurations

Configuration 1 Base case: 2nd supply, backup generator, equipment-level UPS Base case minus equipment-level UPS Base case plus a fast switch Base case plus a static switch Base case plus a fast switch minus some equipment UPS Base case plus facility-level UPS minus local UPS 3 + 0 generators, two utility backups 3 + 1 generators, two utility backups 3 + 2 generators, two utility backups

Failure rate 0 480

MTBF (years) 2 083

Annual equipment cost ($) 70 000

Annual cost per kVA ($) 35

2 3 4 5

16 005 0 060 0 060 0 060

0 062 16 667 16 667 16 667

17 500 91 000 185 000 38 500

9 46 93 19

0 065

15 385

364 000

182

7 8 9

1 440 0 145 0 014

0 694 6 909 73 947

205 120 259 233 313 345

103 130 157

G
12 many local UPS systems (and their maintenance hassle). This configuration reduces costs and substantially improves performance. If even higher performance is needed, it gets expensive very fast. A redundant, standalone distributed generation system can provide higher performance (case 9), so can various redundant UPS systems (not analyzed); both of these are very expensive. C2.3.2.3 Distributed Generation Options Distributed generation is attractive for the hospital because backup generation is present and natural gas is readily available. A natural solution is running the emergency load (the red outlets) continuously from generators, which are isolated from the utility feed supplying the rest of the hospital load. If a generator is lost, a static switch jumpers in the utility supply until another generator starts.

C2.3.3 Summarizing Points on the Case Studies


Each of the case studies had widely differing results and optimal solutions. Some generalities were common: One size doesnt fit all. The optimal solution depends on many factors: facility size, how convenient a second utility feed is, cost of facility interruptions, and the sensitivity and types of facility equipment. The outage cost plus the configuration cost does not represent what some facilities are willing to pay for performance. Some facilities are willing to pay much more for better performance (more redundancy, less chance of failure) than their actual outage costs. Load ride-through enhancements effectively increase performance in many situations for a relatively small cost. The utility-side options are most suitable for large commercial and industrial users. Distributed generation options look potentially viable for many high-performance needs (if the utility is present as a backup).

BIBLIOGRAPHY
[1] Brown R. E., Burke J. J., Managing the risk of performance-based rates. IEEE Transactions on Power Systems, vol. 15, pp. 893898, 2000. [2] Christie, R. D., Statistical classification of major event days in distribution system reliability. IEEE Transactions on Power Delivery, vol. 18, no. 4, pp. 13361341, 2003. [3] Coulter R. T., 2001 Electricity Distribution Price Review Reliability Service Standards. Prepared for the Office of the Regulator-General and Service Standards Working Group, Victoria, Australia, 1999. [4] EEI Reliability Survey, Minutes of the 8th Meeting of the Distribution Committee, 2831 March 1999. [5] EN 50160, Voltage characteristics of electricity supplied by public distribution systems.

G
13
[6] EPRI, Analysis of Extremely Reliable Power Delivery Systems: A Proposal for Development and Application of Security, Quality, Reliability, and Availability (SQRA) Modeling for Optimizing Power System Configurations for the Digital Economy, Palo Alto, CA, 2002. Product ID 051207. [7] EPRI, Global Survey of Regulatory Approaches for Power Quality and Reliability, Palo Alto, CA, 2005. 1008589. [8] EPRI, Reliability Benchmarking Methodology, TR-107937, May 1997. [9] EPRI, Service Quality Index Example Application, Palo Alto, CA, 2006. 1010199. [10] IEEE 1159-1995, Recommended practice for monitoring electric power quality, 1995. [11] IEEE Standard 493-1997, Figure 2-6. [12] IEEE Standard 1366-1998, Trial-Use Guide for Electric Power Distribution Reliability Indices. [13] Settembrini R. C., Fisher J. R., Hudak N. E., Reliability and Quality Comparisons of Electric Power Distribution Systems. IEEE T&D Conference, 1991. [14] Shawn McNulty, Primen, The Cost of Power Disturbances to Industrial and Digital Economy Companies. Report for CEIDS, 2001. [15] Short T. A., Electric Power Distribution Handbook, CRC Press, Boca Raton, FL, 2003. [16] Third Benchmarking Report on Quality of Electricity Supply. Council of European Energy Regulators (CEER), 2005.

Das könnte Ihnen auch gefallen