A Model Unavailability Perspective of Existing Nigerian Township Electricity Distribution

The study aimed to investigate and obtain model values of system unavailability and cost of energy not supplied in existing Nigerian township electricity distributions using a robust representative system. Outage data was obtained from system records for analysis using frequency distribution statistics to identify the causes of outages, and the contribution of each cause of outage to the overall system outage. The cost of energy not supplied (CENS) due to faults, was evaluated at a rate of NGN6 per kWHr. An analysis of the results shows that in a distribution feeder (and system) unavailability is mainly due to load shedding arising and faults; feeder unavailability involving load shedding is 0.25, and 0.1 without load shedding. Compared with standard UA value for well managed systems of less than or 0.01, the obtained unavailability values for a feeder show that downtime management was poor, and the corresponding cost – prohibitive. In view of the regular load shedding on the feeders and poor downtime management, this study recommends a need to obtain model loadflow perspective of township electricity distribution to ascertain load carrying capacity, and the application of distribution automation system or other effective strategies to mitigate downtime.


INTRODUCTION
The electric power distribution system is the most unreliable part of a national power supply system. For example, it contributes up to 81% of all customer outages in the UK [1]. The situation in Nigeria is similar as the distribution system is reported to be the most problematic of the three levels of power supply system. Distribution system unavailability in Nigerian electricity distribution is observed to be poor and aggravated after privatization and deregulation, a measure expected to mitigate electricity and system unavailability problems. This situation brings along with it a new perspective on system unavailability and especially cost of energy not supplied in the Nigerian electricity market; and causes of this unavailability.
Distribution voltages in Nigeria are 33 kV, 11 kV, and 0.4 kV. The most common arrangement for the distribution system is the radial system, in which 11kV feeders emanate from the 33/11 kV injection substation and feed 11/0.4 kV distribution transformers at 0.4 kV load centers. According to [2] and [3], the distribution system has posed a lot of problems to providing economic and reliable power supply in Nigeria due to random nature of growth of cities and poor planning. Generally, poor investment in the power system has been cited as one of the major causes of the poor quality of supply [4]; commensurate system development with township expansions was lacking. According to [5], lots of transformer failures were a major cause of fault outage problems in the distribution sytem. These failures were due to overloading, insulation breakdown, poor maintenance arising from irregular inspection and reconditioning, and the use of inferior and obsolete protection facilities in distribution substations.
Under state owned operation, some studies had shown that the distribution system had poor reliability values [6][7][8]. According to [6] and [7], by deduction, the system unavailability was described as 0.36 and 0.32 respectively. These values were obtained on single 11 kV feeder-network composition in 2008 and 2011 time stages respectively. A review of these are necessary for effective planning after some years of operation under private control and new more-inclusive data. In this study the search for model values and conditions of unavailability and especially CENS is expanded to multiple feeder-networks.
According to [9], the Bureau for Public Enterprises declared that aggregate technical, commercial and collection losses sustained by the various privatized distribution companies were between 40 and 50 percent. This paper provide applicable value based unavailability review, assessment, and cost implication for these emerging distribution companies that have taken over the electricity distribution in Nigeria.
For this study, a test system was selected -11 kV township distribution system of Akure metropolis (7 0 15"00"N, 5 0 11"42"E), the capital of Ondo State in Nigeria. By political plan it became a Millennium City (supported by world agencies) in 2006 hence its current rapid expansion despite existing static capacity limits of power distribution transformers. Figure 1 shows the location of Akure and Ondo State in Nigeria. From a survey in the data period, the radial test system consisted of four 33/11kV substations fed from a 132/33kV transmission substation (with three transfomers). The total capacity of the 33/11kV substations was 70MVA. Figure 2 shows the Akure feeder arrangement.

79
Distribution transformer (DT) population and distribution in the test-network are presented in Figure 3 the DTs feed from seven 33/11kV power transformers, which in turn feed from 3x132/33 power transformers. There are eight 11 kV feeders (Fdrs) or delivery areas, denoted alphabetically as A (Oba-ile fdr), B (Ijapo fdr), C (Alagbaka fdr), D (Ilesha rd fdr), E (Isinkan fdr), F (Ondo fdr), G (Oke eda fdr), and H (Oyemekun fdr). Feeder DT population distribution vary from 25 to 144; DT nominal capacities range from 50kVA to 500kVA (144x500kVA, 146x300kVA, 110x200kVA, 80x100kVA and 14x50kVA); Total installed capacity was 146500kVA. The largest contributor is Feeder C (29%). In 2012, this feeder was alleviated by introducing four more feeders into the delivery area, which were powered by newly installed 33/11 kV power transformers (TFs) located at different areas of the delivery area. These include a 2.5MVA (feeding 7 DTs), and a 7.5MVA TF (feeding two feeders). However, their DT populations are still grouped under their parent feeder C in this study.

Figure 3. No of Distribution Transformers per Feeder
Basic power system reliability concepts discussed in literature as [1], [6], [7], [10] were considered in deriving an effective algorithm for obtaining model values for Nigerian township system in spite of load shedding events and without need for direct consideration of connected customers. The obtained results would enhance justification for investment in planned off-line and on-line network maintenance to obtain quality service, which is also a requirement by regulations [11], [12].

RESEARCH METHOD
From the system records, 288 serial measurements of outage data were obtained simultaneously for each of the following variables on eight feeders: monthly outage events (E) due to load shedding ("LS"), earth fault ("EF"), open-circuit ("OP"), and others ("O") respectively ( Figure 4); and corresponding outage durations ( Figure 5) and power loss during outage. Each serial measurement is obtained for a month. The derived data were further grouped into two categories: events and durations with load shedding and without load shedding. These measurements were considered as data for a dummy feeder, representative of the feeders in the network, since the study is about obtaining values for a non-specific feeder in the network.  The cost of energy not supplied (CENS) on each feeder arising from faults and system failures for the period of the study was evaluated using the following algorithm. From system records outage causes were identified, and the number of outage events and their respective durations for i-th category of outage events in m-th month-measurement were extracted: where and are simple outage event and duration respectively for i-th outage cause, in m-th measurement; LS , EF , OC , PO , O are "load shedding", "earth fault", "open circuit", "planned outage", and "others" categories respectively. Unavailability of feeder in the system for m-th measurement is evaluated as: ; ( where is number of hours in corresponding measurement month, when outage due to load shedding is considered. When outage due to "LS" is not considered, effective number of hours in m-th measurement is considered as : where is outage due to load shedding. Overall outage of system and unavailability in m-th measurement for a feeder: Average unavailability index (AUA) was created and proposed to evaluate feeder UA for measurements: Also 96 measurements correspond to one year data. The energy not supplied due to i-th outage cause for m-th month ( ) is evaluated as: ∑ where and are outage power and duration respectively for e-th outage event, i-th outage cause, and m-th measurement; data of these variables are extracted form system records for the considered period.

81
The distribution system ENS for m-th measurement is obtained as: where and are equivalent active power not supplied (MW) and total outage duration for the system for m-th measurement. Average power loss per feeder for m-th measurement: ENS for M-th interval is defined as: Cost of energy not supplied for period M is evaluated as: where is unit cost of energy, 6 NGN per kWHr. The expected monthly ENS and CENS for period M are denoted , and :

RESULTS AND ANALYSIS
An analysis of outage data reveals the causes of power outages and the frequency at which each factor contributed to power outages in the network on feeder by feeder basis. Unavailability UA on the 11kV feeders comprises the contributions from load shedding (Ls), earth fault (Ef), open circuit (Oc), planned outages (Po) and others (O). Table 1 is a summary of outage frequency and duration distribution statistics of obtained data. Out of 12,961 events, load shedding contributed 66%. Load shedding also accounted for 58% of total outage hours during the study period. Table 1. Summary of Obtained Outage Data Figure 6 shows that, in a feeder, monthly unavailability due to LS could reach 0.54. Whereas UA due to earth fault and OC causes can reach 0.22 and 0.23 respectively. In addition, Figure 7 shows that the monthly unavailability of the system could reach 0.62. Without load shedding, the unavailability of a feeder would reduce to approximately 0.36. In essence, this is the effective unavailability (UA") of a feeder in the system. As M increases to maximum data interval, Figure 8 shows that the average UA of a feeder returns a steadied value of approximately 0.25; and without load shedding, it reduces to 0.10. Average UA of a feeder due to load shedding, earth fault, open circuit faults, planned outage, and others are approximately 0.17, 0.04, 0.03, 0.03, and 0.02 respectively. These values establish that a feeder in the system is in blackout on aggregate period of 36 days annually due to faults and planned outages. Prevailing load shedding increases the expected blackout days to about 91 days. Clearly, the highest contributor to UA is load shedding, and the next in rank is earth fault. This high unavailability values imply that downtime management is substandard. In well run power distribution, UA should be less than 0.01 [1]. Comparatively, the obtained value of 0.1 for the test system is very high and indicates poor downtime management. This finding underscores a need for feeder automation, that is, automated fault location isolation and service restoration (FLISR) system.

FEEDER
Total E  Total_OD, hrs  LS  EF  OC  PO  O  LS  EF  OC  PO  O  A  817  42  103  100  1  1155  257  419  365  0  1063  2196  B  796  125  130  232  11  863  592  472  508  50  1294  2485  C  1001  196  279  437  35  1667  993  1169  1224  106  1948  5159  D  1459  259  181  329  13  6122  1454  801  770  39  2241  9186  E  1634  104  47  104  16  7095  774  326  369  30  1905  8594  F  1604  133  154  177  9  2764  697  577  565  35  2077  4639  G  148  147  163  313  15  517  442  505  519  21  786  2004  H  1064  171  250  152  10  2172  1051  830  490  29  1647  4571  Total  8523  1177  1307  1844  110  22354  6260  5100  4810  311  12961 Figure 9 presents data of obtainable average active power losses in a feeder due to downtime. It shows that the unsupplied MW during downtime can be as high as 11 MW in exceptional cases of LS, but normally clusters around a mean value of approximately 5 MW. Figure 10 shows the ENS obtainable on a feeder for a month, which could be as high as 186,199,754 kWH with consideration for downtime through load shedding. Figure 11 shows that the obtainable ENS due to fault only could only be as high as 26,979,357 kWH, which is a lot lesser in comparison when there is load shedding. Figure 12 shows that the corresponding CENS are approximately 1117 million naira and 162 million naira respectively. Considering varying data interval M, Figure 13 provides obtained cumulative CENS for the data interval; at a scale of 92 serial data corresponding to one year data. Figure 5 shows, by deduction, an average 10 loss of 1,197 million naira per annum due to downtime without load shedding, and an average loss of 16,082 million naira per annum due to downtime with load shedding in a feeder. Following the M concept explained above, Figure 14 provides plots of obtained ACENS for a feeder. Figure 14 shows obtained and corresponding average CENS as 12.52 million naira per feeder per month due to downtime without load shedding and 167.52 million naira per feeder per month with load shedding. From the analyses above, this study provides descriptive values for the distribution system useful for characterizing its prevailing unavailability constraint and reliability, and justifying technical-economical needs for system upgrading schemes, especially automation.

CONCLUSION
Provide a statement that what is expected, as stated in the "Introduction" chapter can ultimately result in "Results and Discussion" chapter, so there is compatibility. Moreover, it can also be added the prospect of the development of research results and application prospects of further studies into the next (based on result and discussion).
In this study an analytical algorithm was deployed to evaluate system unavailability and cost of energy not supplied in a Nigerian township electric power distribution. From an analysis of the results obtained, the following conclusions and contributions are made: i. Distribution Feeder (and system) unavailability is mainly due to load shedding arising (possibly from low generation and insufficient distribution capacity) and faults; feeder unavailability involving load shedding is 0.25, and 0.1 without load shedding. Compared with standard UA value for well managed systems of less than or 0.01, the obtained unavailability values for a feeder show that downtime management was poor. ii. The frequency of planned outages, 1,844 times in 32 months and the high number of hours (4,810 hours) spent on them, an average of 2.6 hours per planned outage, are an indication of a weak or aging infrastructure and a lack of redundancy in the network. iii. The loss of revenue due to poor downtime management was approximately NGN168 million per month (at NGN 6 per kWHr), provided collection rate was adequate and there was no energy theft; approximately, 92% of this amount is due to daily load shedding schemes. In view of these conclusions, it is recommended that model loadflow perspective should be obtained for the township feeders to ascertain capacity to accommodate prevailing load; and downtime mitigating technologies such as automation systems should be installed in the existing electricity distribution infrastructure to mitigate fault detection duration and downtime. A saving, by the utility, of about ten percent of obtained CENS or revenue loss on downtime could fund the implementation of system reinforcements through automation and/or other effective downtime mitigation strategies.