You are here

The flexibility of the RPN approach to model failure and organize maintenance

At the very heart of the RobustPlaNet application domain where Marposs is putting a great effort, there is the well-known issue of trading off hidden and explicit costs of quality production. As clearly depicted in the time diagram of functional level by S. Takata and als. [1], several factors influence the availability of any goods and specifically of production means over their useful life.

Positive maintenance interventions – i.e. those not forced by a breakdown episode – may be inspired by diverse purposes such as time-based, preventive, risk-based, or condition-based policies. The actual economy of a choice is far from being straightforward and the simple common sense may not always play the best advisor. For instance, condition-based prevention may naturally meet the attention of a rigorous planner. However, the capital investment requested for its implementation should be carefully examined against the actual costs of a simpler corrective maintenance policy, especially when maintained components are cheap and quick to replace or their functionality can be predicted with precision (see also [2]).

RobustPlaNet is focusing on very demanding manufacturing scenarios whose complexity, collateral damages and therefore the cost of a breakdown can become highly unpredictable. RobustPlaNet aims at scientifically identifying the minimum set of observable parameters that can guarantee the continuous visibility of the manufacturing system operating conditions. Marposs has thus been cooperating to develop a sensor network capable of acquiring, aligning, elaborating and storing the data necessary to trigger a plurality of maintenance policies (predictive, opportunistic and aggressive). The availability of a choice allows to match the response of the monitored equipment against the production needs of the larger manufacturing systems where the equipment operates.

After about 18 months of work, the methods applied to design the sensor network and data collection have already proven to be open and flexible. In fact, it has been possible to deal with two significantly different manufacturing stages, a multi-axis milling machine for the rough cutting of large automotive engine blocks and a multi-axis, sophisticated flexible manufacturing system engaged in cutting titanium blocks for the aeronautic industry.

[1] Maintenance: Changing Role in Life Cycle Management
S. Takata1 (1), F. Kimura2 (1), F.J.A.M. van Houten3 (1), E. Westkämper4 (1)
M. Shpitalni5 (1), D. Ceglarek6 (2), J. Lee7
1Waseda University, Japan, 2The University of Tokyo, Japan,
3University of Twente, the Netherlands, 4Fraunhofer IPA, Germany, 5Technion, Israel,
6University of Wisconsin-Madison, USA, 7University of Wisconsin-Milwaukee, USA

[2] Product Modelling for Model-Based Maintenance
F. J. A. M. van Houten’ (l), T. Tomiyamaz, 0. W. Salomons’
1 Laboratory of Production and Design Engineering, University of Twente, Netherlands
2 Department of Precision Machinery Engineering, University of Tokyo, Japan

Powered by Drupal