Abstract

This research project proposes a new methodology and tool to perform ongoing commissioning of central plants. The proposed methodology includes a new approach for the development and use of benchmarking models in the context of ongoing commissioning. Different techniques are explored to establish the benchmarking models: (1) a static approach, which is based on pre-defined training set size and established different models for week days and weekend & holidays, or (2) window techniques, which are either augmented or sliding. Two different types of benchmark models are evaluated: correlation-based and Artificial Neural Network (ANN) models.

The proposed ongoing commissioning methodology is evaluated for two chillers installed in the central plant of the Concordia Sciences Building (CSB). Both chillers have identical capacity and performance characteristics; however, they have quite different operating hours. The results show that models developed with seven days of data monitored at the beginning of the summer season provide accurate results over the remaining of the summer and for the following summer. For the chillers used in the case study, the proposed multivariable polynomial (MP) models provide the most accurate prediction with CV(RMSE) below 7% over the remaining of the summer season, and below 8% for the following summer season.

As part of the ongoing commissioning approach, measured data used to develop the benchmarking models combined with manufacturer’s information were also used to develop a calibrated computer model of the CSB central cooling plant in TRNSYS. User input files were modified to reflect the operating characteristics of the equipment installed in the central plant and a control equation was proposed for the cooling towers. The simulation results were in good agreement with the monitored data, with CV(RMSE) that do not exceed 5.5% for water temperature at key locations, 12.5% for the electric power input of the cooling equipment, and 18.6% for the COP of chillers and various groups of equipment. The Relative Error (R.E.) calculated over the summer season for the cooling electricity used is within ±15.6%.

The approach undertaken to calibrate the CSB central cooling plant showed that it is possible to develop a calibrated model using measurements already available from the Monitoring and Data Acquisition System (MDAS) and manufacturer data, without modifying by trial-and-error some variables or using stochastic approaches.