Big data, analytics and the Internet of Buildings

Big data, analytics and the Internet of Buildings

Disclaimer: The views presented in this article are personal views of the author and in no way represent those of the company he works for or any industry body he is associated with.

Buildings have always been rich sources of data. Limitations in technology kept the data stored in native or simple aggregated repositories with some basic analytical functions performed on them. High storage costs were an impediment to growth of such data and analytics. Data and analytics remained limited to subject of interest and enquiry because commercial models around data were under developed. This reduced the meaningful use of data.

For the past several decades there have been many initiatives to develop fault detection and diagnostics and apply huge volumes of building data for improved predictive maintenance. This in a way recognized that there is intrinsic value in data. Most of these initiatives remained in the academic realm due to prohibitive costs, complexities and scalability issues. Another set of initiatives over the years have targeted benchmarking and improving building performance using data driven analytics; these also experienced limited success and limited mass adoption.

There are several contributing factors why the adoption of Big Data Analytics has been limited in the buildings industry so far:

  • It was very difficult, slow and expensive to collect and store all the data for meaningful analytics.
  • Popular technology choices required large volumes of structured, normalized, and error free data. The diversity of buildings, subsystems in buildings and usage makes such an effort monumental.
  • Building data is always imputed with variety and veracity, causing data to lose usability and normal analytics reporting too many exceptions.
  • Not all building related information was available in a digital data format. Paper records can be scanned and converted to digital but converting them into data formats has been quite challenging without enabling technologies.
  • In absence of more robust commercial models, high volume performance data from buildings and subsystems were used for alarms and event management; relegating the data to play a more operational supportive role.

Advancement in big data & analytics technologies has opened up the prospect of possibilities. With the new capabilities around Big Data Analytics, we can do more around buildings.

Building data is both structured and unstructured. Operational data of building systems is often very structured, but maintenance information is much unstructured. Even structured data sharing some common parameters but emanating from different sources can look very difficult. Normalizing and harmonizing this array of diversity is extremely difficult. Traditional technologies were able to wade through structured data reasonably well, but were unable to adequately make use of the rich unstructured data that existed around buildings, since traditional technologies require normalization for any kind of analytics. Current Big Data technologies also allow us to datafy static textual data like maintenance documents, system specifications, drawings and similar material and use the datafied information further.

Building ecosystems are constantly subjected to various phenomena driven by environmental, usage or electro-mechanical changes. Big Data technologies enable us to datafy and leverage these influences in analytics and subsequent actions.

Previously we always started with a hypothesis around optimal performance of buildings or systems contained therein based on our expertise. Prejudices driven by the expanse of our knowledge and intuition always influenced the starting point of hypothesis. We can now improve effectiveness of predictive modeling with sophisticated machine learning algorithms and natural language processing. Availability of more data now can compensate for any inadequacy in the algorithms which previously required high level of expertise and enormous efforts. Pervasive availability of data and models built around actual data eliminates limitations caused by hypothesis and assumptions in models.

Visualization of data and insight becomes a critical factor as we continue to consume more data and indulge in deeper insights. Powerful visualization tools evolved with internet and Web 2.0 technologies, but they were constrained somewhat in meeting the dynamic and evolving needs of buildings, especially in handling unstructured data. New visualization techniques evolving to meet the demands of Big Data Analytics visualization is significantly unlocking the value of building data.

Previously, most of the decisions were made by people considered experts in various aspects of building design and operations. Usually such people came from mechanical, electrical, civil, and environmental engineering or sciences background. In future, with further proliferation of Big Data technologies in the buildings industry, data scientists will also have an increasing say in decisions around buildings. This will bring fresh perspective to the design and management of buildings.

Big Data Analytics has introduced many new challenges to the buildings industry. Some of the key ones are:

  • Taxonomy definition and upkeep. Buildings are living like and evolving ecosystems. By using data we can get insights. However both data and insights revolve around a set of definitions and taxonomies. Since buildings continue to evolve, data correlation and insights also need to evolve in tandem. This requires the taxonomies to almost evolve in a self learning mode. The industry needs to learn new skills and create guidelines around initial definition of taxonomies related to various aspects of building design and operation, and monitoring and maintaining those definitions.
     
  • Designing integrated delivery ecosystems. In many industries, data analytics can work independent of operations and other business functions using the outcome of data analytics. In buildings, the interplay of various data sources and their impact on each other is very dynamic; this influences data and analytics more significantly than most industries. Companies with leadership aspirations need to build integrated delivery ecosystems for leveraging data and analytics. For example, doing predictive fault modeling based on operational performance of equipment and systems in a building without the accompanying maintenance and service delivery functions limits the utility of the predictive models. Independent insight providers using data and analytics will find limited success and scalability in the buildings industry.
     
  • Commercializing the value of data. In the world of buildings, revenue streams of every ecosystem player are tied to some form of tangible action or material insights that leads to a quantifiable benefit. In many other industries the data itself is a source of revenue and almost can be treated as a product. This phenomenon will permeate into the buildings industry too. The richness of data is very high and it can lead to multiple untapped revenue opportunities. Creating commercial models around data and resolving ownership of data and analytics will be a new challenge for buildings industry.
     
  • Rethinking IT infrastructure. Companies participating in the buildings industry have built their IT infrastructure around transactional financial and operational needs. Those are not architected to take advantage of the huge streaming and static data that is collected through various equipment and systems installed in buildings. For example utility companies have sophisticated billing analytics systems, however they may not have toolsets or capabilities to do deeper analytics on how power is consumed by different equipment and systems in the buildings or even to correlate occupancy, weather inputs and set points of different HVAC equipment to modulate temperature, humidity and power consumption. HVAC and mechanical service providers have lot of rich maintenance records. Now they are also developing capabilities of capturing their systems’ performance data, but they do not have the capabilities to modulate their service offerings and schedules dynamically using all this data. Without the right technology infrastructure that combines non-traditional Big Data Analytics with conventional existing IT systems, the usage of data will be limited by human imagination and intervention to a larger than required level.

In less than 20 years, the Internet has progressed to the Internet of Things and changed the world. Big Data Analytics will enable the future Internet of Buildings, where each building will be connected to other buildings. Buildings will interact with peer buildings without human intervention. Buildings will self correct themselves based on environmental, operational and occupational factors.

The prospect of possibilities has just opened up. There is an exciting journey ahead.

Image by Dragana Gerasimoski via Shutterstock