Experimental validation of an energy model of a day surgery/procedure centre in Victoria

Highlights

• Short-term monitoring and sub-metering to quantify energy consumption are applied.

• An overview is presented on recent calibration methods.

• Calibration process using a detailed simulation model is presented.

• Energy audit identified areas where energy is wasted.

• Simulation on certain critical variables is performed.

Abstract

The healthcare sector encompasses a wide range of hospitals and healthcare facilities. As a result of their various sizes and functions, they have different patterns of energy consumption. The number of day surgery/procedure centres is steadily increasing in Australia. With advances in technology, almost eighty per cent of the number of operations that used to be performed in acute care facilities are now done on an outpatient basis, that is, day surgery. It is necessary to provide realistic energy performance benchmarks that are attainable for each typology within this sector. Building performance simulation (BPS) is a powerful and efficient tool to achieve this target. A first step to establish a simulation-based energy benchmarks is to create an accurate simulation model. Validation of the model will ensure it matches the actual building performance. This paper details a method of developing the building energy simulation model for an existing free-standing day surgery/procedure centre that is located within the campus of the main general hospital in Geelong, Australia. In the second part of this paper, simulation runs are performed and investigates the model potential and energy consumption repercussions as a result of tuning certain input parameters. In the case study building, the energy used for electricity, heating, cooling and air conditioning is supplied from various sources. For this reason, this paper uses data gathered from various resources: energy audits, short-term monitoring, and reviews on energy consumption in different typologies and heat transfer formulas to estimate the total energy consumption of the building and establish an accurate simulation model in the absence of whole energy consumption data. Results show a good agreement between the breakdown of the annual energy consumption of the actual building and the energy model. Analysis was conducted using the validated model on some parameters that were found to be inefficient during the building assessment to investigate their impact on the energy performance and indoor environmental quality.

Introduction

Climate change and the increase in the levels of greenhouse gases in the atmosphere are faster than prediction models developed [1]. In Australia, non-residential buildings account for approximately 8% of the total final energy consumption with an annual growth of 2.8% in 2013–2014 [2]. Sustainable and energy-efficient buildings can play an important role in reducing the carbon emissions accompanied by this growth. In recent years, interest has increased in developing tools and methods to forecast and predict the energy performance of buildings to evaluate the potential energy savings of different energy conservation measures (ECM). A recent study by Coakley et al. [3] showed that in general, two types of models are used for predicting building energy performance, namely law-driven models such as dynamic thermal simulation that are based on physics and heat transfer principles, and data-driven models (inverse models). The latter models are developed based on previous or historical data. These data can be either measured at the building site, obtained from energy bills and/or simulation models. The “training dataset” i.e. historical data are then analysed and used in estimating the energy performance of the building. In order to analyse these data, three approaches were identified: black-box, grey-box and detailed model calibration [3].

Currently, the widely-accepted methods when a model deemed to be validated is ASHRAE Guideline 14 which specifies an accepted error range for monthly or annual energy consumption such as Mean Bias Error (MBE) and Coefficient of Variation of Root Mean Squared Errors (CvRMSE) [4]. During the past few years, data-driven approaches such as regression models, artificial intelligence and machine learning have gained popularity over numerical detailed simulation in research as they provide accurate predictions [5]. The reasons for that preference are that detailed simulation models have complex structures, longer development times and need greater computational resources [6], [7]. However, if the simulation models are established with high-quality data, they provide detailed and highly accurate energy forecasting for buildings [3]. One drawback of the data-driven models is that they can only predict the performance of pre-defined input parameters because in predicting building performance they rely on historical data, finding patterns and relationships in those data. This means they are not able, for instance, to evaluate the impact of building orientation or some passive design strategies such as adding overhangs or fins to the building envelope unless the historical data for those parameters is obtained from iterative simulation such as in Lam et al. [8]. Over the years, building performance simulation has matured and become a reliable tool for the energy performance assessment of buildings and energy-efficient buildings design decision-making [9]. However, the initial model in various studies has often shown large discrepancies from the actual building [10], [11]. Validation of these models are necessary steps to ensure an accurate representation of the actual buildings. In regards to validation, methods are classified as either manual or automated. As such, two approaches were identified, which can be employed in both methods: analytical tools and mathematical/statistical techniques [3]. The most common reason found for the discrepancy in performance between the simulated and actual building is the uncertainty in the input parameters in the simulation model and wrong assumptions postulated on the occupants’ schedule [12]. A typical approach to decrease this uncertainty is to conduct sensitivity analyses on the input parameters by changing their values one at a time and running simulations iteratively to identify the most influential ones [5], [13]. Following that, fine-tuning for these parameters is performed until the performance gap is reduced to an acceptable range. Calibration is an indeterminate problem and no agreement has been reached on an accepted standardised method for calibration procedures that can be applied on various building types [14].

Although there are a few studies on evaluating and benchmarking the energy performance of healthcare buildings, the focus of those studies was mainly on healthcare centres and hospitals with limited or almost no studies on day surgery/procedure centres. Further, for those studies on hospitals, the focus usually is on a functional area such as wards [15] or on a particular energy system [16], [17]. Literature on the assessment of day surgery/procedure centres was lowest in number even though they are considered to have the highest energy use intensity among the healthcare facilities. There is still a need to focus on the energy performance of small healthcare facilities and understand their usage patterns and the factors influencing energy consumption, which are quite different from large-size hospitals. In Australia, the number of day procedure centres is growing rapidly [18]. To date, no energy benchmarks have been published for this sector. Due to an increased number of day procedure centres, the development of energy performance benchmarks is imperative in order to set up energy reduction targets for both new design and existing buildings retrofits. In Australia, the National Australian Built Environment Rating System (NABERS), a performance-based rating system, does not encompass healthcare buildings within its portfolio. In addition, whole building simulation models with detailed zones, which are used in this paper, have rarely been deployed for complex buildings, particularly in Australian context. As mentioned before, it is a time-consuming process and requires a high level of accuracy and modelling skills. Hence, such models are avoided by many researchers, and instead statistical models are used to reduce the number of influential parameters and find the best-fit model.

The objective of this study is to develop a validated building simulation model for a healthcare facility i.e. a day procedure centre in Victoria that can act as a base model for the benchmarking process. This paper presents the methods implemented and steps followed to attain this objective. The study also reports on the process employed to quantify the energy consumption profile of the building, and the barriers and challenges that have been encountered during the calibration process. Simulation was run on certain input parameters and energy consumption repercussions are investigated.