Analytics

Detailed specification of available analytics functions and their application.

Available analysis functions

Application notes

  • Unit sensitivity: To this state, our algorithms are unit sensitive. Every pin and attribute is specified with an unit. Mind the specifications.

If unit conventions are disregarded, this can lead to errors and even misleading results of algorithms.

  • Improving results: We continuously improve our analyses functions, interpretations and recommendations. Thus, performing exactly the same analysis at a later point in time can lead to different results.

Example Analysis

This example leads through our description of analytics functions by providing exemplary values and descriptions. The first paragraph of an analysis function provides a short introduction to the analysis.

Quick Start
Description
Results
Example
Components
Application
Quick Start

Value

In other words: Why should I spend time to dig deeper into this analysis.

  • Quick entry into analytics function specifications

  • Exemplary template for analysis function specifications and documentation structure

Recommended for component types

  • All components in need for an example

Description

The Example Analysis is an example on how we structure the documentation of our analysis functions. The goal is to show you where to expect which kind of information to make your work as effective as possible.

In general the description provides information about the concepts of the analysis function, its use-case scenario, and value.

Results

Qualitative warning level

Enum representations of traffic light colors providing a quick overview over the analysis result.

Enum

Color

0

green

1

yellow

2

red

Interpretations

Interpretation text of the condition analysed.

Recommendations

Recommendation texts on how to improve the operation of the analysed component.

KPIs

Providing deeper insights to the cycle behavior. KPIs support human reasoning.

Example KPI

Brief description of the example KPI, its determination, application, and value.

KPI Identifier

Value Range

Unit

example.kpi

0 to 1

kW

Another Example KPI

...

Example

The Example tab provides a hands-on example for each analysis function.

The example building is equipped with an Example Component. This setup is used as a real test bench for the Example Analysis function...

In general you can expect a short demonstration on how we applied the analysis during our development and which results we got from our test bench.

Components

Pins

  • example pin

  • ...

Most of the analysis functions require a mapping for the same pins regardless of the component data model. These pins are listed here. Component-related variations, exceptions, and deeper insights are provided component-wise further down this chapter.

Attributes

  • example attribute

  • ...

Some of the analysis functions require attributes regardless of the component data model. These attributes are listed here. Component-related variations, exceptions, and deeper insights are provided component-wise further down this chapter.

Components

The short text snippets below a component provide information about how to utilize the analysis function for this component. In case you are new to component data models, we recommend to read the high-level introduction to components and check out the component data models for deeper insights.

...

Application

The Application tab offers further information for real application of the analysis.

Recommended Time Span

4 weeks

The Recommended Time Span gives the recommended time span for which the analysis function is intended. The time span is oriented to the time constant of the phenomenon investigated, e.g. in order to analyse weekly accruing patterns, a time span of 4 weeks is recommended. In shorter periods of time, the analysed phenomenon might not occur, while longer periods might lead to superposition of other effects such as seasonal influences and thus blurring of the analysis results. Of course, analysis functions can be used over flexible periods of time at your personal discretion.

Recommended Repetition

every 3 months

The Recommended Repetition provides a recommendation when the analysis should be repeated. The recommendation refers to a continuous operation of the analysed system. In the case of changes to the system, both physical and operational, a repetition of the analysis is generally recommended. Of course, analysis can be repeated at your personal discretion.

Another recommendation

...

Control Loop Oscillation Analysis

Documentation currently under construction 🚧

Heat Flux Analysis

Documentation currently under construction 🚧

Heating Curve Analysis

Documentation currently under construction 🚧

Operating Cycle Analysis

The Operating Cycle Analysis investigates and interprets the cycle behavior of components. Besides identifying sub-optimal cycle behavior, the algorithm provides recommendations on how to improve cycle rates, and KPIs for deeper insides.

Quick Start
Description
Results
Example
Components
Application
Quick Start

Value

  • Lower operating costs

  • Higher energy efficiency

  • Longer equipment and component life times

  • Smoother system integration

Recommended for component types

  • Energy conversion plants

  • Components with high start-up expenses

  • Heat pumps

  • Combined heat and power

  • etc.

Description

The Operating Cycle Analysis aims towards a more constant operation of components and thereby towards a reduction of operational costs. Frequent start and stop processes lead to energy losses and higher wear and tear of the component compared to a constant operation. Further, a frequently alternating operation of a component, e.g. a heat pump, has negative effects on adjacent components, which are enforced to alternate as well. At the same time, too low cycle rates are an indication of a possible under-supply.

The analysis function identifies excessive as well as too low cycle rates on the basis of the datapoint operating message of the analysed component. It provides interpretations of the current cycle behavior and recommendations on how to improve it. The recommended measures include adaptions of control parameter, simple structural measures, as well as instructions for diagnosis and adjustment of the cause of sub-optimal cycle behavior. Additionally to the automated generation of interpretations and recommendations, the algorithm offers a set of KPIs. They provide deeper insides in the cycle behavior of the analysed component.

Results

Qualitative warning level

Enum representations of traffic light colors providing a quick overview over the operation quality regarding cycle rates.

Enum

Color

0

green

1

yellow

2

red

Interpretations

Interpretation text of the cycle rates of the analysed component in regard to its overall operation time.

Recommendations

Recommendation texts on how to improve cycle rates of the analysed component including adjustment of control parameter as well as recommendations on how to investigate the deeper cause for unpleasant cycle behavior.

KPIs

Providing deeper insights to the cycle behavior. KPIs support human reasoning.

Operating time

Operating times KPIs provide information on the total time of operation of the analysed component during the analysed time span.

KPI Identifier

Description

Value Range

Unit

operating time

Total time of operation

0 to inf

h

operating time.relative

Total time of operation divided by total time span

0 to 100

%

Start

The Start KPI is the count of starts of the analysed component during the analysed time span.

KPI Identifier

Description

Value Range

Unit

starts

Count of starts

0 to inf

count

Closed operating cycle

A closed operating cycle is defined as period of time between a start nin_iof the component an the next start ni+1n_{i+1}. The KPI closed operating cycles represents the number of cycles observed during the analysed time span used to determine the operating cycle KPIs (cycle times, duty times, switch-off times).

KPI Identifier

Description

Value Range

Unit

closed operating cycles

Count of closed operating cycles

0 to inf

count

Cycle times

Cycle time KPIs evaluate the cycle times of the closed cycles observed during the analysed time span. The mean, time-weighted average, minimum, and maximum cycle period are determined. In case there were no closed operating cycles observed during the analysed time span, none of the KPI variables are returned on API call.

KPI Identifier

Description

Value Range

Unit

cycle times.median

Median of cycle periods

0 to inf

h

cycle times.average

Time-weighted average of cycle periods

0 to inf

h

cycle times.maximum

Longest cycle period

0 to inf

h

cycle times.minimum

Shortest cycle period

0 to inf

h

Duty times

Duty time KPIs evaluate the duty times of the closed cycles observed during the analysed time span. Duty time is defined as the time of component operation in a closed cycle. The mean, time-weighted average, minimum, and maximum duty period are determined. In case there were no closed operating cycles observed during the analysed time span, none of the KPI variables are returned on API call.

KPI Identifier

Description

Value Range

Unit

duty times.median

Median of duty periods

0 to inf

h

duty times.average

Time-weighted average of duty periods

0 to inf

h

duty times.maximum

Longest duty period

0 to inf

h

duty times.minimum

Shortest duty period

0 to inf

h

Switch-off times

Switch-off time KPIs evaluate the shutdown times of the closed cycles observed during the analysed time span. Switch-off time is defined as the time of component shutdown in a closed cycle. The mean, time-weighted average, minimum, and maximum switch-off period are determined. In case there were no closed operating cycles observed during the analysed time span, none of the KPI variables are returned on API call.

KPI Identifier

Description

Value Range

Unit

switch-off times.median

Median of switch-off periods

0 to inf

h

switch-off times.average

Time-weighted average of switch-off periods

0 to inf

h

switch-off times.maximum

Longest switch-off period

0 to inf

h

switch-off times.minimum

Shortest switch-off period

0 to inf

h

Example

The Operating Cycle Analysis was applied to a real test bench, the heat pump of the E.ON Energy Research Center, RWTH Aachen University. Thus, a heat pump component model was instanced and the respective datapoint mapped to the pin operating message. Figure 1 shows the time series recorded for an exemplary period of 6 hours on a winter day.

Figure 1: Plot of operating message time series from 12:00 am till 6:00 pm

The short shut-down times between periods of duty are conspicuous. This is a hint for unnecessary frequent start and stop processes of the heat pump, leading not only to energy losses but also to high wear and tear of the heat pumps scroll compressor.

In order to apply the Operating Cycle Analysis for the instanced and mapped heat pump, we defined an analysis config via the aedifion API. Executing the Operating Cycle Analysis for the same period of time as figure 1 returns following results:

Qualitative warning level: red (as in traffic light signal color)

Interpretation: Highly increased number of start procedures/operation cycles. A reduction of the starting procedures or a more continuous operation is recommended.

Recommendations:

  • Check whether the system flow temperature runs into the safety limiter and causes safety shutdowns.

  • Check whether the heat requirement is covered too quickly by the producer. If yes, check the modulation operation of the system and, if available, the integration of the buffer tank and the interconnected operation of several generation systems.

  • Check the installation of a power choke. By reducing the electrical power consumed, the amount of heat provided can be reduced and thus the effects of an oversized plant. Pay attention to the manufacturer's specifications as to whether the installation of a choke is technically possible.

KPIs:

KPI

Value

Unit

operating time

4.31

h

operating time.relative

71.8

%

starts

16

count

closed operating cycles

15

count

cycle times.median

0.36

h

cycle times.average

0.369

h

cycle times.maximum

0.41

h

cycle times.minimum

0.34

h

duty times.median

0.254

h

duty times.average

0.264

h

duty times.maximum

0.304

h

duty times.minimum

0.234

h

switch-off times.median

0.105

h

switch-off times.average

0.106

h

switch-off times.maximum

0.106

h

switch-off times.minimum

0.105

h

The automated interpretation confirms our visual analysis of the time series shown in figure 1, summed up by the qualitative warning level "red". The recommendations provide further instruction on how to isolate and fix the cause for the increased number of start and stop processes.

Further, the result offers an advanced set of KPIs, providing additional insights into the cycle behavior of the heat pump. They support human reasoning for case-by-case analysis. For example, the similarity of switch-off times per cycle are peculiar. Since a coincidence would be surprising for such a dynamically operated plant, one could reason, that the heat pump is permanently requested by the higher level controller, but runs into a condition causing a component shutdown for a predefined period of time. Deeper human investigation of the condition is possible via data visualization of aedifion.io and the combination to insights of other analysis functions.

Components

Pins

  • operating message

Components

Boiler

Fan

Results are limited to KPIs.

Heat Pump

Application

Recommended Time Span

1 - 3 days

  • Mind weekends

Recommended Repetition

Every months

  • Cycle rates have a strong seasonal effect

  • Frequent repetition allows to identify operational bad points

Outside Air Temperature Sensor Analysis

Documentation currently under construction 🚧

Room Air Quality Analysis

Documentation currently under construction 🚧

Schedule Analysis

Documentation currently under construction 🚧

Setpoint Deviation Analysis

Documentation currently under construction 🚧

Temperature Spread Analysis

Documentation currently under construction 🚧

Information

The library of analytics functions is constantly expanded. If you are missing an analytics function, wish to implement your own functions, or want us to implement it for you, feel free to contact us.