CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0590
MAINTENANCE SERVICE MANAGEMENT SYSTEM TO IMPLEMENT STRATEGIES TO IMPROVE SERVICE QUALITY H. O. HENRIQUES
W S. ANDRADE
E. P. TAVARES
C. E. MOTTA
S. L. CAPARROZ
UFF University - Brazil
CEFET/RJ - Brazil
UFF University - Brazil
AES - Eletropaulo
AES - Eletropaulo
[email protected]
[email protected]
professoregberto@yahoo .com.br
[email protected]
[email protected]
ABSTRACT This paper describes a Maintenance Service Management System developed for AES Eletropaulo, the utility that supplies energy to São Paulo city in Brazil. A fuzzy inference system is developed to evaluate performance of contractors and in house personnel. Also, specific maintenance works made by contracted and proper staff are evaluated to generate actions to improve performance of the services. This paper shows that energy supply can have the reliability improved with the application of a Maintenance Service Management System to supervise maintenance services.
INTRODUCTION São Paulo is a state located in the south-eastern region of Brazil. AES Eletropaulo, ELPA, is the utility that supplies 38% of the energy consumed in this state, representing 11% of the total Brazil’s consumption. The company serves 5.7 millions customer accounts within a service area of 4.526 km2. The power system consists of 141 Substations, 1700 km of Transmission Lines, 1720 distribution feeders, consisting of 17000 km of overhead primary feeders, 18000 km of overhead secondary networks and 8000 km of underground networks, besides over 180000 distribution transformers. These equipments require 850 emergency maintenance services per day in a total of 8544 work orders and an average of 25.000 inspections per month. To handle this job, ELPA counts on 800 crews in which 40% of them are from electrical maintenance service contractor companies. This maintenance policy is adopted to streamline the system operation, but the effectiveness must be measured. It is known that the maintenance work of a crew influences in the annual reliability indices and a bad service causes, in a short period of time, outages in the system. This paper describes a Maintenance Service Management System, MSMS, developed for AES Eletropaulo, to show how reliability performance of the electrical system is strongly dependent on how the maintenance is handled, either by proper or contracted staff. MSMS were developed in two main modules. The first has the objective of assess the ELPA and electrical maintenance service contractor companies’ performance. The second module assesses the services of a company with low performance indices to determine what has to be improved.
CIRED2009 Session 6
Paper No 0590
The fuzzy inference system developed for the assessment process and the results of its application in a pilot area is described in the next sections.
FUZZY INFERENCE – A BRIEF RESUME This section will present the foundations of Fuzzy systems [2,3] necessary to understand the process developed to assess maintenance services made either for contractor companies or by ELPA’s crews. A membership function (MF) is a curve that defines how each point in the input space is mapped to a membership value (or degree of membership) between 0 and 1. The simplest membership functions are formed using straight lines. Of these, one of the simplest is the triangular membership function (trimf), and it is nothing more than a collection of three points forming a triangle. The trapezoidal (trapmf) is another very popular membership function. It has a flat top and really is just a truncated triangle curve. These straight line membership functions have the advantage of simplicity and for this reason have been chosen to be used in the assessment process[4]. The next step in establishing a system of fuzzy logic capable to assess maintenance service is to define what operations will be used . In this case, EQUAL, COMPLEMENT (NOT), UNION (OR), and INTERSECTION (AND). In order to do this rigorously, we must state some formal definitions [5]: 1: Let X be some set of objects, with elements noted as x. Thus, X = {x}. For example: Maintenance_ Indices = {Productivity_coeficient, SAIDI, SAIFI, etc.}. 2: A fuzzy set A in X is characterized by a membership function mA(x) which maps each point in X onto the real interval [0.0, 1.0]. As mA(x) approaches 1.0, the "grade of membership" of x in A increases. For example mLow(Productivity_ Coefficient) = trapmf(0.0 0.0 0.2 0.4). 4: Complement : mA' = 1 - mA. 5: C = A UNION B, where: mC(x) = MAX(mA(x), mB(x)). 6: C = A INTERSECTION B where: mC(x) = MIN(mA(x), mB(x)). Fuzzy sets and fuzzy operators are the subjects and verbs of fuzzy logic [5]. The if-then rule statements are used to formulate the conditional statements that comprise fuzzy logic. A single fuzzy if-then rule assumes the form “if x is
CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0590
A then y is B” where A and B are linguistic values [5] defined by fuzzy sets on the ranges (universes of discourse) X and Y, respectively. The if-part of the rule "x is A" is called the antecedent or premise, while the then-part of the rule "y is B" is called the consequent or conclusion. An example of such a rule might be “If the productivity coefficient is high and the annual interruption per customer is low then the contractor is good”. Fuzzy inference [1] is the process of formulating the mapping from a given input to an output using fuzzy logic and is described by the following steps[6]:
Emergency_ Indices = {PC, SAIDI, SAIFI, NCE LV, NCE MV, MDT, MTTR};
Step 1 - Fuzzify Inputs :The first step is to take the inputs and determine the degree to which they belong to each of the appropriate fuzzy sets via membership functions.
SAIDI – System Average Interruption Duration Index;
Step 2 - Apply Fuzzy Operator: If the antecedent of a given rule has more than one part, the fuzzy operator is applied to obtain one number that represents the result of the antecedent for that rule. Step 3 - Apply Implication Method: The consequent is reshaped using a function associated with the antecedent . The input for the implication process is a single number given by the antecedent, and the output is a fuzzy set. Implication is implemented for each rule. Two built-in methods are supported, and they are the same functions that are used by the AND method: min (minimum), which truncates the output fuzzy set, and prod (product), which scales the output fuzzy set. Step 4 - Aggregate All Outputs: Aggregation is the process by which the fuzzy sets that represent the outputs of each rule are combined into a single fuzzy set. The input of the aggregation process is the list of truncated output functions returned by the implication process for each rule. The output of the aggregation process is one fuzzy set for each output variable. Step 5 - Defuzzify: The input for the defuzzification process is a fuzzy set (the aggregate output fuzzy set) and the output, however, is a single number and so must be defuzzified in order to resolve a single output value from the set. The next section will describe how the process of assessment of the contractor´s companies was developed.
Where: PC –Productivity Coefficient; LH i PC ; i SLH i LH – Labor hours effectively worked by the crew in service i; SLH – Standard labor hours expected for the service i;
SAIFI - System Average Interruption Frequency Index; NCE LV - Number of emitted work orders for low voltage emergency services; NCE MV – Number of emitted work orders for medium voltage emergency services; MDT – Mean down time; MTTR – Mean time to repair. Each index as a fuzzy variable has the following terms: low, medium and high. The figure 2, shows the process diagram. The memberships functions used for rating performance is shown in figure 3. Twenty four rules have been created to perform Mandani inference. A few of them are listed in table 1.
PC
SAIDI MANDANI : : : :
Step 2,3
PERFORMANCE
Step 4,5
MTTR
Step 1
Figure 2. Performance’s process structure diagram
MODULE I – COMPANIES ASSESSMENT The main objective of this module is to determine assessment fuzzy rates to qualify companies, using the results of a Fuzzy Inference Method, based on some indices collected after the execution of services in feeders under their responsibility, personnel accident and training rates. The performance assessment is done considering two groups of services: maintenance and emergency. The sets of indices considered to judge performance are:
Figure 3. Memberships function for the fuzzy Performance.
Maintenance_ Indices = {PC, SAIDI, SAIFI, NCE LV, NCE MV};
The inference of the first line (rule) of table 1 is done the following way: If PC is low and SAIDI is high and SAIFI
CIRED2009 Session 6
Paper No 0590
Insufficient
Regular
Good
Excellent
CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0590-
is high and NCE LV is high and NCE MV is high Then PER is Low. Table 1. A few rules to infer the performance PC
SAI DI H H M L
L M H H
SAI FI H L L L
NCE LV H L M L
NCE MV H M L L
PER I R G E
The final qualification of the company is given by another inference method that takes into account indices related to personnel qualifications, safety and accidents prevention. Figure 4 shows the process diagram. The indices that will be used as input are: PER – Performance index; AR – Accident rate, defined as the ratio between the number of accidents and the total number of services made by a company during a period of time. If, although, during this period of time has occurred a fatal accident, AR assumed the value of 1; TRI – Training index, defined as the ratio between the number of hours the employees attended training programs and the number of hours employees should attend training programs according to ELPA standards, during a period of time; Memberships functions for the fuzzy Qualification are the same as the performance, showed in figure 3. Twenty seven rules have been created to perform Mandani inference and a few of them are listed in table 2. As a result, a score is produced in the step 5. The defuzzification method developed was the centroid. Timely and accurate performance assessments will identify factors hindering contract success.
MODULE II – SERVICE ASSESSMENT
This module measures, in annual basis, the maintenance and emergency services quality of the feeders in a similar way of module I. A group of 14 services have been chosen for overhead and 6 for underground feeders. For each type of service and for each company, a research was made to survey data of how much times the service has been done in a period of time (FRE), its mean time to repair (MTTR) and the Productivity Coefficient (PC) of the single service. These three input data are used by the inference method to assess those services. Figure 5 shows the process. The memberships functions used for rating the service are the same of those shown in figure 3, for performance. Twenty seven rules have been created to perform Mandani inference and a few of them are listed in table 3. The next section describes some results of the assessment of two contractors responsible for the maintenance of eight feeders.
FRE MANDANI MTTR
Step 2,3
SERVICE RATE
Step 4,5
PC
Step 1
Figure 5. Fuzzy rating of maintenance or emergency service. Table 3. A few rules to infer Service FRE H M L L
MTTR L M M L
PC L M H H
SERVICE Insufficient Regular Good Excellent
PER
PRACTICAL APPLICATION RESULTS
MANDANI Step 2,3
AR
Step 2,3
QUALIFICATION
Step 4,5
TRI
Step 1
Figure 4. Fuzzy rating qualification process. .Table 2. A few rules to infer Qualification PER L M H H
AR H M L L
TRI L L M H
E – Excellent; G – Good; R – Regular; I – Insufficient; TEC_IND – Technical index.
QUA Insufficient Regular Good Excellent
Corrective action plans may be utilized to get the contract back on track. ELPA deals with 600 contracts per year. If a company had a regular or insufficient qualification, it will have their main services assessed to point out what has to be done to correct its performance and qualification. The next module describes how to do this. CIRED2009 Session 6
MSMS has been implemented in an area where the maintenance services are performed by SOCREL CO, which is responsible for the feeders VAR0103, LUB0115, PRE0107, LUB0113 and TOBACE CO, responsible for TSE0103, LAP103, JAG0108, BAL0108. Table 4, shows the performance assessed for the year of 2007, where :
Paper No 0590
According to the assessment process, the contractor with rate R and I must have their services verified. Table 4 shows that Feeders JAG 0108 (R), LUB 0113 (I), LUB 0115 (R), PRE 0107 (I), were selected for further verification using module II. Table5 shows the final qualification of the contractors according ELPA quality standards rules. Table 4. Performance assessment, year 2007.
CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0590-
Co TOBACE TOBACE TOBACE SOCREL SOCREL SOCREL TOBACE SOCREL
Feeder NCEMT NCEBT SAIDI BAL 0108 4,40 5,00 6,53 JAG 0108 1,60 3,90 25,76 LAP 0103 2,20 2,20 4,17 LUB 0113 9,80 33,00 25,54 LUB 0115 12,60 29,80 9,50 PRE 0107 20,90 4,00 43,55 TSE 0103 3,80 3,30 8,72 VAR 0103 8,00 31,80 12,98
SAIFI 4,11 7,58 3,44 14,83 4,08 20,03 3,23 4,77
PC PER TEC_IND 0,80 E 8,22 0,81 R 5,00 0,66 G 7,65 0,63 I 2,27 5,00 1,00 R 0,53 I 1,90 7,00 0,50 G 0,65 R 5,00
Table 5. Contractors assessment Co TECN_IND AR TRI(%) QUA SOCREL 3,54 0,7 111,74 I TOBACE 6,97 0 121,26 R
Table 6 shows the assessment of three services executed very often in these feeders by SOCREL. They have been coded 715, to represent the replacement of screw connector, 746, related to replacement of low voltage assembled cable and 446, repair of low voltage triplex cable. Table 6. Services assessment Code 715 746 446
PC 0,67 0,29 0,10
MTTR 45 60 51
FRE 357 204 215
Service R I I
TEC_IND 4,96 1,88 1,83
After corrective actions have been made, the assessment for the year 2008 is shown in table 7.
advisable to survey occurrences, in a past data history, to eliminate the ones that never occurs. The results of implementation have showed that the reliability indices have improved after the detection and correction of the wrong procedures of the crew in the field, pointed out by module II of MSMS. The observation of the contractor’s maintenance services with low TEC_IND by the ELPA’s specialists made possible the correction of the flaws in the maintenance process and, the sensation of being controlled, made the contractors improve other services too. It is also possible to rank the contractors by its technical index and, in the near future, to rank them according to the services skills. Other very important issue is that, in MSMS, the training and safety of the team have the same weigh as the technical index. MSMS proved to be a strong tool to control contractors performance and to build a strategy to improve service quality. REFERENCES
[1]
Table 7. Performance assessment, year 2008 Co TOBACE TOBACE TOBACE SOCREL SOCREL SOCREL TOBACE TOBACE SOCREL
Feeder NCEMT NCEBT SAIDI BAL 0108 0,67 0,57 2,09 JAG 0108 1,18 0,55 0,82 LAP 0103 1,30 0,75 1,00 LUB 0113 1,65 0,66 6,45 LUB 0115 1,37 0,48 8,18 PRE 0107 3,09 0,87 8,45 TSE 0103 1,08 0,76 1,55 TSE 0111 1,07 0,86 4,36 VAR 0103 0,69 0,25 3,45
SAIFI 1,64 0,36 1,36 3,91 6,00 14,45 1,91 2,09 1,27
PC PER TEC_IND 1,00 E 9,20 0,92 E 9,20 0,64 G 7,45 0,93 E 9,20 0,86 R 5,00 0,76 R 5,00 0,50 G 7,00 0,83 E 9,20 0,86 E 9,20
Table 7 shows improvement of performance not only in the feeders rated I and R in 2007, but also in others of the same contractor. The overall technical index has also improved. Table 8 shows this index at the end of year 2008.
[2]
[3] [4]
[5]
Table 8 Technical Index for year 2009. Co
TECN_IND
SOCREL
7,1
TOBACE
8,41
[6]
CONCLUSIONS
This paper presented an application of Fuzzy Logic to assess contractor’s maintenance and emergency services. One of the most important results conceived in this work is the possibility to avoid opposite subjective opinions on a single issue, during services assessment. The assessment rules, with the use of MSMS, are equal for everyone in ELPA, though a few may not totally agree with them. It is important to gather the most maintenance experienced employees and build the rules with them. This implementation has showed also that the availability and the quality of the input data is essential for the entire process. Depending on the number of input variables, the number of possible rules may be very large. For this reason is
CIRED2009 Session 6
Paper No 0590
E.H. Mamdani,, "Applications of fuzzy logic to approximate reasoning using linguistic synthesis," IEEE Transactions on Computers, Vol. 26, No. 12, pp. 1182-1191, 1977. H.J. Zimmerman, Fuzzy Set Theory and its Applications, 3rd edition, Kluwer, Boston, 1996. L.A. Zadeh, "Fuzzy Logic," Computer, Vol. 1, No. 4, pp. 83-93, 1988. L.A. Zadeh, "Knowledge representation in fuzzy logic," IEEE Transactions on Knowledge and Data Engineering, Vol. 1, pp. 89-100, 1989 L.A. Zadeh, "The concept of a linguistic variable and its application to approximate reasoning, Parts 1, 2, and 3," Information Sciences, 1975, 8:199-249, 8:301-357, 9:43-80 T.J. Ross, 1995, Fuzzy Logic With Engineering Applications, MacGraw-Hill, NY, USA, pp. 371402.
