^{1}

^{2}

^{3}

In this article we are discussing the suggestion of planning maintenance in such situation that we have failures in database of failures and maintenance registered also according to type of failure— priority or non-priority. We are assuming that maintenance must be mainly oriented on avoiding major failures of machines. The effect of maintenance manifests with delay and is dependent on development of failures. We express this problem as linear programming task that is solvable with Solver-extension of application Excel. We demonstrate the suggested solution on example of failures of machines between years 2010 and 2012.

In the paper [

Excel and its supplement Solver present also suitable tools for solving our problem in this case. Range of the problem allows its solution in real time. We will demonstrate our solution on illustrated example with these following parameters: we have 285 records of machine failures, of which 203 records are records about the way of machine failure. We need triplet of data (YEAR, MONTH, and LOSS) to create the input data, similarly as in the model [

But specifying priority of failures may be a problem. One of the possibilities is to create a frequency table from failure code list, which represents the number of occurrences of the considered failure in watched period. Then we can sort this table in descending order and declare the first x failures as a priority. This approach has a logical reasoning as we preferably try to look after those machines that show increased rate of failure [

We’ve created a table of frequencies of failures in single months and in years 2010-2012 and based on it we have created graphs of the development of priority failures (

By summarizing the above mentioned frequencies we get a table that allows us to determine the exact number of maintenance which is done in each month and their impacts on the identified priority and non-priority failures. We have part of the data in table (

More informative than the table is the following chart of development of failures and maintenance. We see that preventive maintenance avoids failures relatively poorly and certainly not prioritized ones (

The following model tries to remove the above present disadvantage. Let pi_{1} priority failures and pi_{0} non- priority failure be identified during n-periods and it u_{i} maintenance schedules (

We are looking for such plan of maintenance schedules x_{i}, where the sum of coefficient of efficiency from maintenance priority machines y_{i} and coefficient of efficiency from maintenance from non-priority machines is maximal [

Variables are integer values x_{i} in this model, that indicate the number of planned maintenance in the i-th period, which reflects preference of maintenance cause of priority failure. Coefficient of maintenance efficiency y_{1} can be interpreted as a percentage from the number of failures, that have to be covered by priority mainten- ance in that same month.

We are getting to the following task of linear programming:

Total number of failures and maintenance schedules | |||||
---|---|---|---|---|---|

Year/month | Year | Month | Priority failure | Non-priority failure | Maintenance |

01.2010 | 2010 | 1 | 0 | 2 | 3 |

02.2010 | 2010 | 2 | 0 | 1 | 4 |

03.2010 | 2010 | 3 | 3 | 0 | 6 |

04.2010 | 2010 | 4 | 0 | 3 | 2 |

05.2010 | 2010 | 5 | 5 | 0 | 3 |

06.2010 | 2010 | 6 | 1 | 8 | 2 |

07.2010 | 2010 | 7 | 0 | 1 | 0 |

08.2010 | 2010 | 8 | 0 | 2 | 1 |

09.2010 | 2010 | 9 | 2 | 1 | 1 |

10.2010 | 2010 | 10 | 1 | 3 | 5 |

11.2010 | 2010 | 11 | 1 | 8 | 2 |

12.2010 | 2010 | 12 | 1 | 6 | 2 |

01.2011 | 2011 | 1 | 5 | 0 | 5 |

02.2011 | 2011 | 2 | 1 | 16 | 5 |

03.2011 | 2011 | 3 | 4 | 0 | 2 |

04.2011 | 2011 | 4 | 0 | 7 | 5 |

The target function “Equation (1)” maximizes total coefficient of maintenance efficiency. Restrictive condition “Equation (2)” ensures that the number of preventive maintenance in the cycle is greater than the number of already performed maintenance. Restriction “Equation (3)” ensures that the number of non-priority failures in the current period will be covered by efficiency taken from preventive maintenance in the previous and current periods. Restriction “Equation (4)” provides the same coverage of non-priority failures only at the current period. Cycles in periods are formulated by restriction “Equation (5)” Conditions “Equation (6)” and “Equation (7)”are forced by character of variables [

Before using the Excel Solver it is necessary to define the cells of the worksheet on the basis of conditions “Equation (1)”-“Equation (7)”of the model (

Results of this solution are shown in (

We are getting y_{0} = 0.31 and y_{1} = 0.50 which leads to an overall average efficiency of 40.62%.

Computer experiment with the presented model “Equation (1)”-“Equation (7)” has shown its applicability to real data. Solver of linear programming in Excel allows optimal scheduling of preventive maintenance that will be used for the benefit of preventing the priority failures as much as possible. Moreover it offers also a quantita-

_i | p_i1 | p_i0 | u_i | x_i | |
---|---|---|---|---|---|

Month/year | Period | Priority | Non-priority | Maintenance | Maintenance schedules |

01.2010 | 1 | 0 | 2 | 3 | 3,00 |

02.2010 | 2 | 0 | 1 | 4 | 2,00 |

03.2010 | 3 | 3 | 0 | 6 | 2,00 |

04.2010 | 4 | 0 | 3 | 2 | 2,00 |

05.2010 | 5 | 5 | 0 | 3 | 3,00 |

06.2010 | 6 | 1 | 8 | 2 | 2,00 |

07.2010 | 7 | 0 | 1 | 0 | 2,00 |

08.2010 | 8 | 0 | 2 | 1 | 3,00 |

09.2010 | 9 | 2 | 1 | 1 | 3,00 |

10.2010 | 10 | 1 | 3 | 5 | 2,00 |

11.2010 | 11 | 1 | 8 | 2 | 3,00 |

12.2010 | 12 | 1 | 6 | 2 | 1,00 |

01.2011 | 13 | 5 | 0 | 5 | 3,00 |

02.2011 | 14 | 1 | 16 | 5 | 2,00 |

03.2011 | 15 | 4 | 0 | 2 | 2,00 |

04.2011 | 16 | 0 | 7 | 5 | 2,00 |

05.2011 | 17 | 1 | 7 | 1 | 3,00 |

06.2011 | 18 | 3 | 9 | 2 | 2,00 |

07.2011 | 19 | 3 | 7 | 0 | 2,00 |

08.2011 | 20 | 1 | 4 | 0 | 3,00 |

09.2011 | 21 | 1 | 4 | 0 | 3,00 |

10.2011 | 22 | 4 | 12 | 1 | 2,00 |

11.2011 | 23 | 2 | 4 | 2 | 3,00 |

12.2011 | 24 | 2 | 12 | 2 | 1,00 |

tive computing the effectiveness of these maintenances. We believe that if we will know also other parameters, for example average expense for maintenance, removal priority resp. non-priority failures or total financial source we will be able to generalize this more in the economic aspect. We will see limitations of the presented model in larger size of data that Excel can no longer process. Then it is appropriate to use other software to optimize, for example Gurobi optimization. For practical use of our proposed model it is necessary to determine more priorities.

This publication was supported by the European social fund within the framework of realizing the project “Support of inter-sectoral mobility and quality enhancement of research teams at Czech Technical University in Prague”, CZ.1.07/2.3.00/30.0034 and the contribution has been elaborated within the task solution: “E-learning of robotics with the implementation of virtual laboratory with remote control of real facilities on based Internet”.

ErikPrada,AlenaPešková,MichaelValášek, (2015) Model of Maintenance Planning Based on Trend of Machines Failures with Two Priorities. World Journal of Engineering and Technology,03,205-210. doi: 10.4236/wjet.2015.34021