Cost effectiveness of computerized laboratory automation

The objective of this investigation was 1) to develop a methodology for determining potential productivity increases and cost savings to be realized from computerized automation of laboratory instruments, and 2) to evaluate the cost effectiveness of increased automation in the IBM San Jose Research Laboratory, using the developed methodology. The various factors are discussed which were considered in arriving at an algorithm for calculating cost and productivity for both manually operated and automated instruments. The algorithm is defined explicitly. The results of a survey using this methodology are presented. The survey covered 75 experiments involving 57 instruments, a number of which have been connected to a computer for several years. The instruments considered include gas-liquid and gel permeation chromatographs, mass spectrometers, IR and optical spectrometers, micro-densitometers, electron spectrometers (Auger and microprobe), X-ray spectrometers, thermal analyzers, NMR spectrometers, miscellaneous instruments, and a variety of custom-built experimental apparatus. The distribution of cost for various instrument categories is presented. Automation cost and the underlying assumptions are discussed in detail. The particular implementation considered in the survey is based on a sensor-based computer (IBM System/7) which is shared by several experiments and in turn linked to a general-purpose host computer (IBM 360) for progrmn preparation and extensive data analysis. The results obtained indicate significant cost savings for about 80 percent of the experiments considered. The remaining 20 percent for which automation could be justified only on the basis of increased produtivity are also dissussed. The distribution of automation costs points up the relatively high cost associated with the deveolpment of applicatiom programs. The fairly consistent distribution of cost for a wide variety of experiments suggests that the cost savings and the calculated increase in productivity due to automation could be expected in other multi-instrument laboratories.