1. Introduction
Ground source heat pumps (GSHP) are highly efficient new and renewable energy systems consisting of heat pump units and ground heat exchangers and are used for cooling, heating, and generating hot water [
1]. The occurrence of faults during the operation of heat pump units can make GSHPs highly inefficient and energy-intensive. Faults such as refrigerant leak, fouling, compressor leak, and reduction in secondary fluid flow rates (SFFR) can reduce heat pump performance by about 30% [
2,
3,
4]. This has attracted great research interest in heat pump fault detection and diagnosis (FDD) in recent times.
Many researchers have proposed FDD models for single faults occurring in heat pumps. Choi et al. Reference [
5] developed an FDD model for refrigerant charge faults in a ground source heat pump using the degree of subcooling since it is directly related to the refrigerant charge amount. Payne et al. Reference [
6] used a data-clustering methodology to detect and diagnose faults in packaged air-conditioners. The methodology was developed to apply laboratory-based FDD algorithms to that of real systems installed on the field. Gasche et al. [
7] modeled a two-phase FDD algorithm for refrigerant faults in rolling piston compressors to determine if temperature affects the extent of refrigerant charge faults. The study found that refrigerant faults predicted using isothermal FDD models differ greatly from those predicted using non-isothermal FDD models. Cho et al. [
8] developed correlations for the cooling mode performance parameters of a heat pump imposed with single faults according to indoor and outdoor temperature conditions. Yoo et al. [
9] developed a methodology to detect slow and long-term refrigerant faults in a residential air conditioner using the difference between the inlet air temperature and midpoint temperature of a heat exchanger. The study found that the trend of the temperature difference was not affected by varying outdoor temperature conditions. Wang et al. [
10] developed an FDD model for various heating, ventilation, and air conditioning(HVAC) subsystems with sensor faults and found that various faults in heat pump systems can be detected using processed data from sensor measurements. Casteleiro-Roca et al. [
11] developed a new FDD approach for geothermal heat exchanger faults using classification techniques. Saththasivam et al. [
12] developed an FDD mechanism for common faults in chillers using the standard thermodynamic model. The model was used to obtain thermal resistance and internal entropy generation as a coefficient, which was used to detect condenser fouling and flow rate faults. Sellami et al. [
13] developed an FDD model to detect single faults in a refrigerator compartment using the bond graph method, which was done using linear fractional transformations. The authors adopted the bond graph approach due to its merit as far as the implementation of processed data is concerned. Zhao et al. [
14] proposed an FDD algorithm for centrifugal chillers and tested it experimentally on fouling in the condenser, refrigerant faults, and un-condensable gas faults. Noel et al. [
15] experimentally studied the impacts of refrigerant charge faults and heat exchanger fouling on the performance parameters of a variable speed compressor heat pump.
A few studies have also been carried on heat pump FDD for multiple faults. Zhao et al. [
16] developed a decoupling-based FDD model to bridge the gap between laboratory and real-time FDD applications. The FDD model detected single and multiple-simultaneous refrigerant faults in a chiller. Han et al. [
17] used a combined support vector machine and multi-label methods to automatically detect and diagnose faults in a building chiller. The study had enough experimental data to validate the single FDD model but had limited experimental data to validate the multiple-simultaneous FDD model. Du and Jin [
18] used Fisher discriminant analysis to develop FDD algorithms for multiple-simultaneous faults in air handling units. The faults considered in the study included water valve and sensor faults.
According to literature, there is a decrease in the secondary fluid flow rate due to the blocking of strainer in heat pumps [
19]. However, most works available in the open literature on heat pump FDD have focused on single faults that occur in the refrigerant side with little research on secondary fluid flow rate fault detection in the open literature. The few multi-fault FDD models available in the open literature have focused on sensor faults and faults occurring at the refrigerant side [
16,
17,
18]. Meanwhile, multiple faults occurring at the secondary fluid side of heat pumps is inevitable. This study, therefore, analyzes the effect of simultaneously occurring condenser and evaporator secondary fluid flow rate faults on the performance of a water-to-water heat pump unit. The study also develops a methodology to detect single secondary fluid flow rate faults (SSFF) and multiple-simultaneous secondary fluid flow rate faults (MSSFF) for the water-to-water heat pump using linear correlations and rule-based fault categorization tables. The FDD methodology uses the measurement from temperature sensors to detect faults in heat pumps. It is therefore cheap and easy to apply to the control systems of heat pumps for the early detection of single and multiple-simultaneous secondary fluid faults to prevent performance reduction, higher energy consumption, and higher operating costs of heat pumps.
2. Materials and Methods
The test rig used for this study is presented in
Figure 1. It was equipped with a compressor, condenser, electronic expansion device (EEV), evaporator, and a four-way valve. It is similar to the experimental setup used in the work of Boahen et al. [
2] to develop an FDD model for refrigerant charge faults in heat pump units. In this study, the use of the test rig focused on analyzing the effect of multiple-simultaneous secondary fluid flow rate faults on the performance of heat pumps and developing FDD methodology for single and multiple-simultaneous secondary fluid flow rate faults.
Figure 2 shows a picture of the experimental setup, where 1 is the condenser, 2 shows the evaporator, 3 shows the compressor, 4 shows the expansion device, 5 and 6 show the evaporator and condenser constant temperature water baths respectively, which forms the secondary fluid flow loop, 7 is the data acquisition unit, 8 is the power meter used in measuring the compressor power consumption and 9 shows the computer used for data storage. The test rig had a refrigerant flow loop and secondary fluid flow loops. R410A was used as the refrigerant, while brine of 40% ethylene glycol concentration was used as the secondary fluid in the secondary fluid flow loops. The test rig had inverter-driven pumps and a constant temperature water bath in the secondary fluid flow loops to simulate the evaporator and condenser secondary fluid flow rate faults.
The tests were conducted in cooling mode with the evaporator and condenser acting as the indoor heat exchanger (IDHX) and the outdoor heat exchanger (ODHX) respectively. During operation, R410A is compressed into vapor refrigerant which rejects heat to the secondary fluid in the condenser and gets condensed to a subcooled state. The subcooled refrigerant is expanded by the EEV to become a refrigerant of low pressure and temperature. The expanded refrigerant absorbs heat from the brine in the evaporator to turn superheated and is then compressed in the compressor into refrigerant of high pressure and temperature for the cycle to continue.
The first step in the experimental process was to determine the reference conditions at the standard cooling mode inlet water temperature of 25 °C and 12 °C across the condenser and evaporator respectively, according to ISO 13256-2 [
20]. The reference test parameters were found as 4700 g optimum refrigerant charge, degree of superheat of 7 °C, and 8 LPM brine flow rate across the evaporator and condenser. The reference condenser and evaporator secondary fluid flow rates (SFFR) were designated as 100%. After establishing the reference test parameters, the evaporator SFFR was varied at 60%, 80%, 100%, 120%, and 140% of the reference value at constant reference condenser SFFR to simulate evaporator SFFR faults. Afterward, the condenser SFFR was varied at 60%, 80%, 100%, 120%, and 140% of the reference value at constant reference evaporator SFFR to simulate condenser SFFR faults. The evaporator SFFR and condenser SFFR were then varied simultaneously at 60%, 80%, 100%, 120%, and 140% of the reference values to simulate multiple-simultaneous secondary fluid flow rate faults (MSSFF), as shown in
Table 1. The cases considered for SSFF include secondary fluid overflow or secondary fluid underflow at fixed reference conditions, while cases for MSSFF include secondary fluid underflow or overflow at the condenser side combined simultaneously with secondary fluid underflow or overflow at the evaporator side, as shown in
Table 2. Secondary fluid flow rates at 60% and 80% of the reference value were considered as underflow, while secondary fluid flow rates at 120% and 140% of the reference value were considered as overflow. All imposed faults were examined at varying condenser inlet water temperatures (
) of 20 °C, 25 °C, 30 °C, and 35 °C to evaluate the effect of
on the SFFR faults. Each experiment was repeated three times in the pre-test to ensure the repeatability and reliability of the collected data. The experiments were controlled by adjusting the EEV opening to achieve a 7 °C superheat.
Sensors were used on the test rig to measure the operating parameters and performance of the heat pump during the test period. Resistance temperature detector(RTD) sensors and volumetric flow meters were installed to measure temperature and SFFR respectively. Thermocouples, pressure transducers, power meter, and mass flow meter were used to measure refrigerant temperatures, refrigerant pressure, compressor power, and mass flow rate respectively.
Table 3 shows the sensor accuracies in the test rig.
Data from the test rig were collected and saved on the computer using Yokogawa MX 100 at 30 min saving time with 3 s scanning time. Cooling capacity (
Q) was calculated as the product of the density of the secondary fluid, the specific heat capacity of the secondary fluid, volumetric flow rate of the secondary fluid in the evaporator, and temperature difference of the secondary fluid across the evaporator, as shown in Equation (1), where
is the density of the secondary fluid,
is specific heat capacity of the secondary fluid.
is the volumetric flow rate of the secondary fluid,
LWT is the temperature of secondary fluid leaving the evaporator, and
EWT is the temperature of the secondary fluid entering the evaporator.
COP was calculated as the ratio of the cooling capacity to power consumption of the compressor, as shown in Equation (2), where
Q is the cooling capacity and
W is the power consumption of the compressor. The uncertainty analysis on the parameters of the heat pump was done using the uncertainty theorem according to the root of the sum of squares of the measured variables [
21], as presented in Equation (3), where
x is the deviation of all measured parameters of the calculated value caused by sensor errors and
U is the uncertainty of the calculated value. The
COP and cooling capacity had uncertainties of 3.1% and 2.9% respectively.
4. Discussion
4.1. Evaporator Secondary Fluid Flow Rate Fault
The cooling capacity of the heat pump according to the evaporator SFFR decreased at underflow conditions and increased at overflow conditions because refrigerant mass flow rate decreased at underflow conditions and increased at overflow conditions, respectively. The rate of heat transfer increased at underflow conditions and decreased at overflow conditions, however, the increasing refrigerant flow rate was higher than the rate of heat transfer at both underflow and overflow conditions. Furthermore, the cooling capacity decreased as increased in all SFFR faults. A decrease in therefore has a similar effect on the cooling capacity as evaporator SFFR overflow. Therefore, for space cooling, the underflow of secondary fluid flow rate in the evaporator will decrease the cooling effect of the heat pump such that the heat pump will not be able to meet the room cooling set temperature at the standard operating condition, while the overflow of the secondary fluid flow rate will increase the cooling effect of the heat pump. Furthermore, COP was not affected by evaporator SFFR faults because COP is affected by the heat pump’s cooling capacity and power consumption. Power consumption slightly decreased at underflow conditions and slightly increased at overflow conditions. This resulted in the COP trend with variation in the evaporator SFFR. Furthermore, the COP decreased greatly with an increase in because of decreased cooling capacity and increased power consumption as increased. COP is related to the operating cost of the heat pump. The results of the study, therefore, show that evaporator secondary fluid flow rate faults have no significant effect on the operating cost of the heat pump unit.
4.2. Condenser Secondary Fluid Flow Rate Fault
The slight decrease in cooling capacity at underflow conditions and the slight increase in cooling capacity at overflow conditions is due to a slight increase in evaporating temperature at underflow conditions and a slight decrease in evaporating temperature at overflow conditions. This means that the heat pump will generate a little lower space temperature below the setting value at underflow conditions and a space cooling temperature slightly higher than the set temperature when the condenser secondary fluid flow rate is higher than the rated value. Furthermore, the compressor power consumption of the heat pump increased at underflow conditions and decreased at overflow conditions. The COP trend was therefore caused by the combined effect of the cooling capacity and power consumption at underflow and overflow faults. This means that when in operation, much energy will be needed for the heat pump to produce space cooling at the set temperature when the condenser secondary fluid flow rate is below the rated value. This will increase the operating cost of the heat pump. The COP decreased by 10% and 21% at 80% and 60% underflow faults respectively and increased by 6.4% and 16.6% at 120% and 140% overflow faults respectively at the reference .
4.3. Multiple Simultaneous Secondary Fluid Flow Rate Fault (MSSFF)
COP decreased at SUUF and SOUF conditions but increased at SUOF and SOOF conditions due to an increase in power consumption at SUUF and SOUF conditions and decrease in the same at SUOF and SOOF conditions. This implies that higher energy consumption and operating costs will be needed for the heat pump to produce the required room cooling setting temperature when the evaporator and condenser secondary fluid flow rates decrease simultaneously below the rated value, or when there is a simultaneous increase in the evaporator secondary fluid flow rate and decrease in condenser secondary fluid flow rate. Furthermore, cooling capacity decreased at SUUF and SUOF conditions and increased at SOOF and SOUF conditions due to a decrease in the refrigerant mass flow rate at SUUF and SUOF conditions and an increase in the refrigerant mass flow rate in the evaporator at SOOF and SOUF conditions. Thus, a simultaneous decrease in the evaporator and condenser secondary fluid flow rates below the rated value and simultaneous decrease of the evaporator secondary fluid flow rate with an increase in condenser secondary fluid flow rate above the rated value will result in the inability of the heat pump to generate the required set temperature of the room to produce the required cooling effect in the room. Therefore, for MSSFF, the cooling capacity is highly dependent on the evaporator SFFR fault, while COP is highly dependent on the condenser SFFR fault.
6. Conclusions
Multi-simultaneous faults result largely in performance reduction and higher energy consumption of heat pumps. However, research on multiple-simultaneous secondary fluid faults is lacking in the open literature. This study discussed the performance characteristics of a water-to-water heat pump according to secondary fluid flow rate faults and developed FDD methodology for a single secondary fluid flow rate fault (SSFF) and multiple-simultaneous secondary fluid flow rate faults (MSSFF) occurring in the water-to-water heat pump unit at varying outdoor temperatures in cooling mode. The SSFF included IDHX underflow, IDHX overflow, ODHX underflow, and ODHX overflow faults, while the MSSFF included simultaneous IDHX underflow and ODHX underflow (SUUF), simultaneous IDHX underflow and ODHX overflow (SUOF), simultaneous IDHX overflow and ODHX underflow (SOUF) and simultaneous IDHX overflow and ODHX overflow (SOOF). The cooling capacity decreased at IDHX underflow, ODHX underflow, and all MSSFF conditions except SOOF. COP increased at IDHX overflow, ODHX overflow, SUOF, and SOOF conditions. However, COP was not affected by IDHX underflow conditions.
A FDD model was developed for the secondary fluid flow rate faults in the water-to-water heat pump unit using correlations and a fault categorization table. The correlations were developed using multiple linear regression for the SSFF and MSSFF. The evaporating temperature, condensing temperature, compressor discharge temperature, and secondary fluid temperature difference across the IDHX and ODHX were used as independent variables to model the secondary fluid flow rate fault correlations. Correlations using the secondary fluid temperature difference across the IDHX and ODHX had the lowest error thresholds and were therefore selected for the secondary fluid flow rate FDD methodology.
The developed FDD correlations were validated by using them to predict the experimentally imposed secondary fluid flow rate faults. For SSFF, correlation using the secondary fluid temperature difference across the IDHX predicted the IDHX secondary fluid flow rate fault within the lowest error threshold of ±1.6%. Correlation using the secondary fluid temperature difference across the ODHX as an independent variable predicted the ODHX secondary fluid flow rate fault within the lowest error threshold of ±2.2%. The MSSFF correlation used the secondary fluid temperature difference across the IDHX and ODHX to predict simultaneous IDHX and ODHX secondary fluid flow rate faults within an error threshold of ±6.4%.
The FDD correlations are developed for the water-to-water heat pump unit used in this study, which has specific component sizes and can therefore not be directly applied to all heat pumps. However, the methodology provides a guide for researchers to develop FDD correlations specific to their systems. Future studies will focus on developing FDD methodologies that can be generally applied in all ground source heat pumps.