Apparatus, computer program product, and method for evaluating course of efficiency degradation in chillers
Methods, apparatuses, and computer program products are disclosed for monitoring chiller efficiency degradation. An example method receives a first data set comprising chiller operations data and chiller efficiency data over a first time interval. The method receives a second data set comprising chiller operations data and chiller efficiency data over a second time interval. The method generates, with the machine learning model, a data prediction based upon the first data set and the chiller operations data of the second data set, wherein the data prediction comprises expected chiller efficiency data over the second time interval. The method determines a chiller efficiency degradation level based on a difference between the data prediction and the chiller efficiency data of the second data set.
1 . An apparatus for determining efficiency degradation of chillers, the apparatus comprising at least one processor and at least one non-transitory memory including computer-coded instructions thereon, the computer-coded instructions configured to, with the at least one processor, cause the apparatus to:
receive a first data set comprising chiller operations data and chiller efficiency data over a first time interval;
receive a second data set comprising chiller operations data and actual chiller efficiency data over a second time interval;
generate, with a machine learning model, a data prediction based upon the first data set and the chiller operations data of the second data set, wherein the data prediction comprises expected chiller efficiency data over the second time interval;
determine a chiller efficiency degradation level based on a difference between the data prediction and the actual chiller efficiency data of the second data set for the second time interval wherein the data prediction comprises expected chiller efficiency data;
determine a predicted coefficient of performance based on the data prediction and a recorded coefficient of performance based on the actual chiller efficiency data of the second data set;
determine a coefficient of performance degradation level based on a difference between the predicted coefficient of performance and the recorded coefficient of performance; and
determine a coefficient of performance prediction error based at least on the coefficient of performance degradation level, wherein the coefficient of performance prediction error is shifted such that its initial value corresponds to a baseline of zero degradation, thereby enabling a normalized degradation profile over time.
2 . The apparatus of claim 1 , wherein the chiller operations data of the first data set and the second data set comprises at least one of temperature of a cooling medium entering a condenser, temperature of the cooling medium leaving the condenser, volumetric flow rate of the cooling medium, mass flow rate of the cooling medium, temperature of a chilled medium entering an evaporator, temperature of the chilled medium leaving the evaporator, volumetric flow rate of the chilled medium, mass flow rate of the chilled medium, and compressor power consumption.
3 . The apparatus of claim 1 , wherein the chiller efficiency data of the first data set, the data prediction, and the second data set comprises at least one of coefficient of performance and power consumption.
4 . The apparatus of claim 1 , wherein the coefficient of performance prediction error describes a coefficient of performance degradation level over time, and wherein the coefficient of performance prediction error is shifted to begin at a level indicating no degradation.
5 . The apparatus of claim 1 , the apparatus further caused to: determine an amount of energy wasted over time based on coefficient of performance degradation level.
6 . The apparatus of claim 5 , the apparatus further caused to:
determine an excess expenditure value time profile based on the amount of energy wasted.
7 . The apparatus of claim 6 , the apparatus further caused to:
determine an optimal maintenance time based on the excess expenditure value time profile and a cost of maintenance.
8 . The apparatus of claim 1 , the apparatus further caused to:
determine the first time interval based at least on a temperature of a cooling medium entering a condenser and a cooling demand load.
9 . The apparatus of claim 1 , the apparatus further caused to:
smooth the first data set, wherein the first data set is smoothed based at least on a third time interval that is longer than a basic sampling time interval and encompasses the basic sampling time interval.
10 . The apparatus of claim 1 , wherein the machine learning model comprises a regression model.
11 . The apparatus of claim 1 , wherein the machine learning model is trained based on the first data set.
12 . A computer-implemented method, comprising:
receiving a first data set comprising chiller operations data and chiller efficiency data over a first time interval;
receiving a second data set comprising chiller operations data and actual chiller efficiency data over a second time interval;
generating, with a machine learning model, a data prediction based upon the first data set and the chiller operations data of the second data set, wherein the data prediction comprises expected chiller efficiency data over the second time interval;
determining a chiller efficiency degradation level based on a difference between the data prediction and the actual chiller efficiency data of the second data set for the second time interval wherein the data prediction comprises expected chiller efficiency data;
determining a predicted coefficient of performance based on the data prediction and a recorded coefficient of performance based on the actual chiller efficiency data of the second data set;
determining a coefficient of performance degradation level based on a difference between the predicted coefficient of performance and the recorded coefficient of performance; and
determining a coefficient of performance prediction error based at least on the coefficient of performance degradation level, wherein the coefficient of performance prediction error is shifted such that its initial value corresponds to a baseline of zero degradation, thereby enabling a normalized degradation profile over time.
13 . The computer-implemented method of claim 12 , wherein the chiller operations data of the first data set and the second data set comprises at least one of temperature of a cooling medium entering a condenser, temperature of the cooling medium leaving the condenser, volumetric flow rate of the cooling medium, mass flow rate of the cooling medium, temperature of a chilled medium entering an evaporator, temperature of the chilled medium leaving the evaporator, volumetric flow rate of the chilled medium, mass flow rate of the chilled medium, and compressor power consumption.
14 . The computer-implemented method of claim 12 , wherein the chiller efficiency data of the first data set, the data prediction, and the second data set comprises at least one of coefficient of performance and power consumption.
15 . The computer-implemented method of claim 12 , wherein the coefficient of performance prediction error describes a coefficient of performance degradation level over time, and wherein the coefficient of performance prediction error is shifted to begin at a level indicating no degradation.
16 . A computer program product comprising at least one non-transitory computer-readable storage medium having computer program code stored thereon that, in execution with at least one processor, is configured for:
receiving a first data set comprising chiller operations data and chiller efficiency data over a first time interval;
receiving a second data set comprising chiller operations data and actual chiller efficiency data over a second time interval;
generating, with a machine learning model, a data prediction based upon the first data set and the chiller operations data of the second data set, wherein the data prediction comprises expected chiller efficiency data over the second time interval;
determining a chiller efficiency degradation level based on a difference between the data prediction and the actual chiller efficiency data of the second data set for the second time interval wherein the data prediction comprises expected chiller efficiency data; and
determining a predicted coefficient of performance based on the data prediction and a recorded coefficient of performance based on the actual chiller efficiency data of the second data set;
determining a coefficient of performance degradation level based on a difference between the predicted coefficient of performance and the recorded coefficient of performance; and
determining a coefficient of performance prediction error based at least on the coefficient of performance degradation level, wherein the coefficient of performance prediction error is shifted such that its initial value corresponds to a baseline of zero degradation, thereby enabling a normalized degradation profile over time.
17 . The computer program product of claim 16 , wherein the chiller operations data of the first data set and the second data set comprises at least one of temperature of a cooling medium entering a condenser, temperature of the cooling medium leaving the condenser, volumetric flow rate of the cooling medium, mass flow rate of the cooling medium, temperature of a chilled medium entering an evaporator, temperature of the chilled medium leaving the evaporator, volumetric flow rate of the chilled medium, mass flow rate of the chilled medium, and compressor power consumption.
18 . The computer program product of claim 16 , wherein the chiller efficiency data of the first data set, the data prediction, and the second data set comprises at least one of coefficient of performance and power consumption.