Often when first being introduced to analytics people look to make comparisons with alarms. After all, doesn’t an alarm programmed in a BAS tell me something is wrong? At a very basic level there is a similarity but if we look just a little bit deeper we see that there are fundamental differences between alarms and analytics.
First of all, alarms require that you fully understood what you wanted to look for at the time you programmed the system. In other words you knew exactly what you wanted to look for and took the time to program that specific definition into the system. This is fine for simple issues like a temperature going outside of a limit. There are many inter-relationships between equipment systems that may not be known at the time the control system was programmed, however. One of the great benefits of analytics is that it enables you to find patterns and issues you weren’t aware of – providing results that show how your building systems are really operating vs how you thought they were operating.
Describing What Matters. The next difference to consider is the capability of expressing what you want to find. Alarming systems don’t typically enable you to implement sophisticated logic that interrelates multiple data items from different data sources. Alarms are typically limit-based checks that act on a single item and perhaps include a minimum time duration before they trigger, but rarely do they implement more sophisticated logic.
A typical alarm might evaluate a single item against a limit at a single point in time – for example, is the KW demand above a certain limit right now? Analytics on the other hand look for patterns across windows of time. Analytic rules crunch through large volumes of time-series historical data to find patterns that are difficult or impossible to see when looking only at real-time data. For example, while an alarm might tell us our building is above a specific KW limit right now, analytics tells us things like how many hours in the last 6 months did we exceed the electrical demand target? And how long were each of those periods of time, what time of the day did they occur and how were those events related to the operation of specific equipment systems, the weather or building usage patterns.
Analytics Goes Beyond Controller Data. Controller-based alarms are also very local in nature. A controller operating an air handler simply can’t combine data feeds from throughout an enterprise so that we can answer questions such as how the energy use correlates with the activity of our business. Analytics enables us to combine data from many different sources beyond just the local controller. This can include enterprise system data such as the revenue generated by a fast food restaurant over a period of time. With this data we can discern relationships or patterns that are important. For example, we would expect more energy to be used in a busy restaurant than in a quiet one. Analytics can quickly find these patterns showing us sites where energy use per dollar of revenue is higher than other sites.
Analytics is an Exploratory Process. Alarms are also not well suited to exploration of data relationships that are a key part of analytics. Findings from one analytic rule often highlight other issues we should look for. An analytics platform makes it easy to explore and test new theories. Attempting to do those types of things in a control system just isn’t feasible. For example, would you be able to justify the cost of reprogramming the alarm logic in 1000 remote sites because you have an idea about a correlation that could be resulting in energy waste and want to test that theory? It is likely you could not justify that expense and the risk associated with modifying and reloading programs.
Large facilities with central plants further highlight this part of the challenge. In facilities with built up HVAC systems, there are a lot of interactions between the various systems that simply aren’t fully understood until after the building is operational. And, these interactions may change over time as building use or other conditions change. A key part of optimizing facility operation is accepting the reality that we are going to discover new things over time. To take advantage of this reality we need tools that enable us to easily find those new patterns and then create analytic rules to automatically watch for those conditions. An analytics system allows you to easily test new ideas and identify new patterns and correlations in the data. You never have to worry about reprogramming a controller.
As we can see, analytics is about more than looking at point values and alarm limits. Analytics finds patterns in data from multiple sources to identify patterns and issues that are important. And that is really the point – analytics are about data relationships – not just data values. Control systems are typically limited in their data manipulation and analysis capabilities – they focus on their internal data items only. You need true data analytics technology to do the job effectively.
An analytics system doesn’t replace or take anything away from the basic alarm functions of a control system – instead it is a natural complement, helping us see what our systems are really doing and why those alarms are occurring.
John Petze Fri 9 Sep 2011
Often when first being introduced to analytics people look to make comparisons with alarms. After all, doesn’t an alarm programmed in a BAS tell me something is wrong? At a very basic level there is a similarity but if we look just a little bit deeper we see that there are fundamental differences between alarms and analytics.
First of all, alarms require that you fully understood what you wanted to look for at the time you programmed the system. In other words you knew exactly what you wanted to look for and took the time to program that specific definition into the system. This is fine for simple issues like a temperature going outside of a limit. There are many inter-relationships between equipment systems that may not be known at the time the control system was programmed, however. One of the great benefits of analytics is that it enables you to find patterns and issues you weren’t aware of – providing results that show how your building systems are really operating vs how you thought they were operating.
Describing What Matters. The next difference to consider is the capability of expressing what you want to find. Alarming systems don’t typically enable you to implement sophisticated logic that interrelates multiple data items from different data sources. Alarms are typically limit-based checks that act on a single item and perhaps include a minimum time duration before they trigger, but rarely do they implement more sophisticated logic.
A typical alarm might evaluate a single item against a limit at a single point in time – for example, is the KW demand above a certain limit right now? Analytics on the other hand look for patterns across windows of time. Analytic rules crunch through large volumes of time-series historical data to find patterns that are difficult or impossible to see when looking only at real-time data. For example, while an alarm might tell us our building is above a specific KW limit right now, analytics tells us things like how many hours in the last 6 months did we exceed the electrical demand target? And how long were each of those periods of time, what time of the day did they occur and how were those events related to the operation of specific equipment systems, the weather or building usage patterns.
Analytics Goes Beyond Controller Data. Controller-based alarms are also very local in nature. A controller operating an air handler simply can’t combine data feeds from throughout an enterprise so that we can answer questions such as how the energy use correlates with the activity of our business. Analytics enables us to combine data from many different sources beyond just the local controller. This can include enterprise system data such as the revenue generated by a fast food restaurant over a period of time. With this data we can discern relationships or patterns that are important. For example, we would expect more energy to be used in a busy restaurant than in a quiet one. Analytics can quickly find these patterns showing us sites where energy use per dollar of revenue is higher than other sites.
Analytics is an Exploratory Process. Alarms are also not well suited to exploration of data relationships that are a key part of analytics. Findings from one analytic rule often highlight other issues we should look for. An analytics platform makes it easy to explore and test new theories. Attempting to do those types of things in a control system just isn’t feasible. For example, would you be able to justify the cost of reprogramming the alarm logic in 1000 remote sites because you have an idea about a correlation that could be resulting in energy waste and want to test that theory? It is likely you could not justify that expense and the risk associated with modifying and reloading programs.
Large facilities with central plants further highlight this part of the challenge. In facilities with built up HVAC systems, there are a lot of interactions between the various systems that simply aren’t fully understood until after the building is operational. And, these interactions may change over time as building use or other conditions change. A key part of optimizing facility operation is accepting the reality that we are going to discover new things over time. To take advantage of this reality we need tools that enable us to easily find those new patterns and then create analytic rules to automatically watch for those conditions. An analytics system allows you to easily test new ideas and identify new patterns and correlations in the data. You never have to worry about reprogramming a controller.
As we can see, analytics is about more than looking at point values and alarm limits. Analytics finds patterns in data from multiple sources to identify patterns and issues that are important. And that is really the point – analytics are about data relationships – not just data values. Control systems are typically limited in their data manipulation and analysis capabilities – they focus on their internal data items only. You need true data analytics technology to do the job effectively.
An analytics system doesn’t replace or take anything away from the basic alarm functions of a control system – instead it is a natural complement, helping us see what our systems are really doing and why those alarms are occurring.