If you want to know how long a boiler will last, where do you start? CIBSE Guide M: Maintenance Engineering and Management is the most widely-used reference document for this topic available in the UK. It provides indicative life expectancies derived from several published sources with additional guidance from a working group of experienced building services engineers.
BSRIA conducted research for the Construction Industry Life Cycle Cost Analysis (CILECCTA 
) European research project. The project aim was to create a user-oriented, knowledge-based software capable of full life-cycle cost analysis. The BSRIA deliverable was the production of life expectancy distribution curves for input to the CILECCTA software.
The difficulty with the CIBSE guidance is that life expectancy is provided as a single year, which can't be used to derive a distribution curve. The advantage of a distribution curve is that it indicates the dispersion of data around the mean, with the mean indicating the most likely life expectancy as opposed to a definitive number.
So what other guidance exists? Chapter 36 of the HVAC Applications Handbook includes life expectancies of mechanical equipment. It is commonly referenced in the US and is the equivalent to CIBSE Guide M. Also, an ASHRAE research project in 2003 developed an interactive web-based database with a specific focus on mechanical equipment service life. The object of the database was to update the life expectancy data in the Handbook (derived from research in 1978) by involving the ASHRAE community to contribute data, and to take into account the developments in technology, materials, manufacturing and maintenance practice.
The data is divided into two sub-sets: replaced, for data indicating installation and removal dates, and currently in service, for data with just an installation date. The data indicates mean and standard deviations for the components, yielding distribution curves. The dataset itself was subsequently analysed, taking year 2010 as the removal date for the currently in service dataset and combining with the 'replaced' data to generate an ASHRAE dataset curve, also shown in the chart.
The only other publication that provides data to generate a distribution curve is Surveyors' Experiences of Buildings in Use, published by the Building Cost Information Service (BCIS). The publication collates the responses from 92 individual building surveyors on the life expectancy of common building components.
Additional sources, including the Kirk & Dell'Isola publication Life Cycle Costing for Design Professionals, provide a single year expectancy charted as a single line (similar to that in Guide M). Other publications, such as The Longevity of Building Services Installations and the Building Services Component Life Manual published by the Swedish Building Research Council collates data from various publications and indicates life expectancy as data ranges.
BSRIA has analysed all of these and then generated a composite curve from the average of the curve data for each component. This was weighted evenly between the sources, and added to the graph as the CILECCTA curve for the software tool.
So, how long will a boiler last? Well, the general overview of the CILECCTA curves of mechanical and electrical components indicate a mean life expectancy of just over 20 years for most of the components. Although many commercial buildings have a design life of 50 years, it is more common for the useful life to be around 25 years. At 25 years, it is likely a major refurbishment would be required to bring a building up to date. Thus the useful life of a commercial building will coincide with the life expectancy of major building service components.
The research also suggests that building services refurbishment should coincide with any refurbishment of the building fabric. This correlation could also assist commercial leasing decisions. If a building's fabric is starting to become dated, this is indicative that the mechanical systems will require additional maintenance and possible replacement of components.
One component that shows a significant deviation on longevity is lighting fixtures. The BCIS data indicates life expectancy around 12 years, while the other data sources indicate a higher life expectancy. Unlike many mechanical and electrical components, light fixtures are highly visible to building occupants, and light levels and distribution are also readily apparent. A possible reason for the shorter life expectancy is not only deterioration of the fixture itself, but also how quickly any deterioration is detected by occupants.
Many of the reference sources reviewed for CILECCTA provide indicative service lives for components, which comply with good practice in design and installation, normal maintenance and typical exposure and usage. As these conditions are unlikely to be replicated exactly in practice, service lives will need to be adjusted to represent the likely in-service conditions. BS ISO 15686 tries to address this by applying a range of factors to a reference life expectancy accounting for quality of components through to in-use conditions.
An area that is not extensively addressed in the study is new technology, including renewable energy systems that currently do not have substantial data for life expectancy. This sector should show significant growth in the future, as the technology becomes more mainstream.
A common term in reliability studies is Mean Time Between Failure (MTBF). This is the predicted time between inherent failures of a system during operation, and a good supplementary indicator of life expectancy. However the definition of MTBF means that it cannot be directly related to life expectancy. This seems counter-intuitive, but is best illustrated by the following example.
A battery may have a five hour life expectancy but a MTBF of 100 000 hours. This implies with a sample of one million batteries operating simultaneously, there will be 10 failures per hour during the expected five hour life of the sample set. MTBF is therefore more closely linked to production, manufacturing and maintenance quality rather than to the inherent life expectancy of a single item in a bespoke installation.
While the research results are interesting, they highlight a lack of datasets for building services components, especially data that accounts for all the various factors addressed in BS ISO 15686. This data has to come directly from owners and operators of facilities. BSRIA is looking to work with its Operation and Maintenance Benchmarking Network to develop its own database.
The long-term aim of the CILECCTA project is to develop robust software for construction industry life cycle cost analysis.
Related reading from theNBS.com
