Abstract:

The paper describes the results achieved by a group of students from Politecnico di Milano technical university during the MSc course “Laboratory of Mechatronics”. The aim of the course is to prepare students to face real working-life problems (e.g. model uncertainties, non-linearities, delays, etc.), to apply their technical and theoretical preparation, and to present their results. The problem presented in this paper is the design of a high performance control logic for an elevator and its implementation on a PLC hardware.
Managing an elevator, setting up its control structure and regulating all the parameters of the system is a task widely studied since it is hard and difficult to be worked out. To design a high performance control strategy, it is necessary to define what kind of building the elevator will serve and subsequently the optimal logic to be adopted. As a matter of fact, a private housing estate will be characterized by a low and sporadic people flow, not constant over time. The elevator of an hospital, instead, will be busy all day long, due to an intense and constant traffic of users.
The aim of this paper is to analyze the strategies to handle the elevator of an office-building, where the workers are expected to use the elevator more likely in some precise moments of the day (morning arrival, lunch-time, end of the working day), called peaks. According to the queuing theory, the arrivals at the floors are described through Poisson processes with specific λ, depending on the considered floor and the time of the day.
Two are the strategies currently used for the control of most elevators: “First-Come–First-Served” (FCFS) and “All-Up-All-Down” (AUAD). FCFS logic serves each call independently serving the calls according to their temporal order; AUAD logic instead gives priority to the direction of the lift movement in order to minimize the overall service time. However, there is not a single logic capable of facing, with the same level of performance, both the peaks and the moments of calm: an innovative algorithm (called “Adaptive”) has been therefore developed in order to take into account the variability of traffic conditions while keeping optimal performances. A number of statistical simulations supports the effectiveness of this adaptive logic.