This paper was produced for the 2019 NAFEMS World Congress in Quebec Canada
In our everyday life lift systems for furniture gain more and more importance, both for kitchen cabinets as well as other types of furniture. Such lift systems, like the ones shown in Figure 1, provide comfortable access to wall cabinets, allowing for a smooth workflow unhindered by cabinet doors.
There is a large variety of different lift systems available, each suitable for a specific application. Various cabinet dimensions, a wide range of door sizes and a large selection of handle bars and materials need to be covered. At the same time, lift systems must be easy to install and convey a pleasant sensation while in use.
Additionally, each lift system has a characteristic motion-sequence which is realized by different numbers of levers and their corresponding pivot locations. Figure 1 shows two lift systems with different opening characteristics. Based on the assumption that the number of levers is known the challenge in designing such a lift system is to find the right position of the pivots such that all requirements are met - the optimum. The design engineer is usually faced with a wide range of possibilities, which leads to numerous parameters with a relatively large design space. Naturally, this flexibility is limited by external requirements that cause additional constraints and thereby limit the feasible positions of the pivots.
Typically, optimization problems of this quality have multiple local optima with only one, the global optimum, being of interest. The existence of local optima means that there are multiple feasible pivot positions that satisfy the constraints. However, between those parameter sets that represent feasible designs there are many other sets leading to a locked kinematic system, meaning that the furniture front cannot be opened at all. If such a lock-up occurs the simulation model is not able to solve and exits with an error. Thus, the solution space can be divided into three groups: Feasible designs, unfeasible designs and unknown designs about which the optimization has no specific information. Naturally, such a discontinuity makes the discovery of feasible designs far more challenging. In addition, usually there are many constraints that cause the feasible area to be very small and hence difficult to find in the first place.
The above makes a proper setup of the optimization problem even more important. Both a good understanding of the system and the provision of correct restrictions are crucial.
This paper intends to show a typical optimization problem within the context of a lift system and presents two different approaches that can assist the setup of an optimization: Firstly, a method to approximate the impact of a parameter on the constraints in a simple way is outlined. Secondly, a strategy to improve the setup of the design-space for individual parameters is suggested. Moreover, the influence of the start population’s size is discussed along with a validation strategy for the obtained results.
|Date||18th June 2019|
|Organisation||Julius Blum GmbH|