Open Collections

The U.B.C. recordings collection : a traffic flow study

University of British Columbia

The U.B.C. Recordings Collection represents a service network with different service centers which include the information desk, the shelves, and the listening room. Arrivals to the system follow a random traffic pattern. The units flowing in the network are defined as jobs requiring service from the service stations. The probabilistic job routings are considered to be a stationary finite-state Markov Chain discrete in time. Field work was done during the summer school and during the regular winter term. The essential traffic observations were obtained by questionnaires. A first analysis of the operating characteristics was made using a graphical method shown in the diagrams 1 and 2; at this, the recordings collection was considered to be a service facility with a large number of service channels. By this method, knowledge could be gained about the number of arrivals during a day, the number of simultaneous customers in the system and in the listening room, the time spent for service, the regularity and changes in arrival and service statistics, and the regularity of departures. It became evident that the distribution of the interarrival times, both to the recordings collection (shelves and information desk) and to the listening room as well, fit a negative exponential distribution function. The distribution of the service times associated with the recordings collection could be identified as exponentially distributed, while the service time distribution of the listening room follows an empirical distribution function (see diagrams 3 to 10). The basically observed statistics were applied in a simulation. The simulation was performed with the "General Purpose Simulation System/360, (GPSS/360)" In three variations in order to obtain empirical data under modified arrival rates and one or two desk servers, respectively. The data of interest was gathered in tables; the information content of the tables was plotted in graphs (see computer output). Data of significance has been used for the description of the traffic behavior of each single service center and was displayed in the diagrams 11 to 13. The flow model predicted that congestion may appear at the information desk and in the listening room. The simulation has confirmed this prediction. Under lasting rush hour conditions, jobs requiring service at the information desk may spend about 3 to 4 times their actual service time in queue waiting for service. This result has been obtained under the general assumption of a first-in first-service order of all arriving desk jobs. However, in practice, the observed rush traffic lasts only about two hours between 12:00 and 2:00. Consequently, arriving jobs related to re-shelving of recordings may be buffered and serviced at times of lower traffic. Thus, the actual waiting time in queue will be lower. In the case of a second server on duty, the waiting time in queue drops down significantly. It will not exceed the service time (see diagram 11). The service at the shelves storing the recordings is problemless. The maximum number of jobs at the shelves does not reach the service capacity. The average service times are constant under varied conditions. In deciding what is an adequate number of turntables, there is a conflict between high utilization rate and satisfied customers. The conflict situation becomes evident under extreme traffic conditions, for instance, when the peak traffic exceeds the provided capacity, or conversely, when the equipment is idle because of low traffic. A satisfying approach depends upon the range and weight of the appearing traffic fluctuations. As it is shown, the intensity of arrivals varies considerably during the interval of a day. However, the range of fluctuating arrival rates still increases by considering different days of a week or different seasons, such as the summer session or the winter term. An acceptable solution of an adequate number of turntables installed may lead to a compromise between 'perfect' equipment utilization and 'completely' satisfied customers. Depending on the pursued policy, it may favour one or the other interest.

Text

2011-04-18

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/33760

Business Administration

University of British Columbia

Graduate