Comparison of a Solar Driven Absorption Chiller and Photovoltaic Compression Chiller Under Different Demand Profiles: Technological, Environmental and Economic Performance ^†

Abstract

HVAC systems are becoming increasingly important around the world due to the increasing need for climatization in recent years. While district heating systems have been used for a long time, district cooling systems tend to be something that is only reserved for large buildings, making decentralized cooling flourish, shaping the idea of considering it as the first choice when it comes to cooling devices, disregarding the efficiency of larger systems. This article compares two technologies for district energy solutions. One option features single-stage absorption chillers using solar thermal technologies (Fresnel collectors) for heat, while the other uses high-efficiency compression chillers with photovoltaic technologies. Parametric studies were used to determine system sizes and considerations were taken to perform such as comparison. This paper concludes that compression chillers are the better option for cooling systems with variable demand while absorption chillers are a good choice for systems with constant demand, like data centers, especially when there is a high-temperature heat source available.

1. Introduction

Residential and building energy demand for HVAC (heating, ventilation, and air conditioning) systems are one of the main targets in the energy decarbonization transition as it accounts for approximately a quarter of the total primary energy consumption in Europe [1,2]. Energy needs in the field are growing organically, being covered with a relevant contribution of non-renewable $C{O}_2$-emitting sources (23% in 2022) [3]. In addition, there is another factor that affects the type of service that is provided by the HVAC systems. As one of the observed impacts of climate change, degree days, as a technical indicator of cooling and heating energy demand [4,5], is being transferred from heating to cooling. The increasing number of cooling degree days (CDDs) implies the increased need for cooling systems. At present, CDD values are becoming difficult to ignore. In Europe, cooling degree days (CDDs) values have been increased from 37 in 1979 to 140 in 2022 [5]. Decarbonization technologies for heating and cooling are traditionally based on different strategies. Heating equipment has been based for ages on devices producing direct thermal heat through boilers, either biomass or fossil-fueled, or solar thermal collectors. Cooling has been mainly fulfilled by compression chillers directly driven by electricity. Such a paradigm is changing with the implementation of heat pumps for both heating and cooling as a highly potential decarbonized option if low-carbon electricity is used.

Different authors have been analyzing the implementation of the available cooling technologies in different networks. Some as Hampo, have taken Life Cycle Assessment (LCA) for compression chillers [6]. Others have taken into account in their research the use of absorption chillers such as Arabkoohsar or Jannatabadi [7,8], the latter with special importance due to the hot climate in which it is implemented. A third more inclusive approach has been made by Sun in his article [9] considering different configurations which include absorption chillers, mainly due to the presence of available waste heat at good temperature levels to be used, and compression chillers as means of storing cold energy. In reference [10], the authors suggest that solar absorption cooling systems could play a significant role in reducing CO₂ emissions linked to the demand for space cooling. However, there is a need to develop control strategies that will allow them to operate effectively under varying conditions of solar radiation and cooling demand. Khliyeva, O. et al. [11] compare heat-driven absorption, ejector refrigerators and electrically-driven vapor compression to evaluate the use of waste heat for cooling. The authors incorporate CO₂ emissions from human labor in the environmental evaluation. The study concludes that the waste heat solution offers better environmental performance than the vapor compression solution, with geography being a key factor in the outcome. In the work of [12], the authors demonstrate that the solar absorption chiller could be a viable solution for supplying the cooling demand of data centers, particularly in arid or semi-arid climates. In the study [13], three different cooling technologies are compared in terms of energy and economic efficiency for meeting the demand of six residential buildings. The technologies examined include district cooling, compression chillers coupled with or uncoupled from photovoltaics (PV) systems, and absorption chillers. The results indicate that the compression chillers with PV systems offer the best technical performance, while district cooling is the most cost-effective option.

For this article, a different approach has been taken, in which two main cases have been analyzed to compare face-to-face these technologies. The first one, a compression chiller electrically supplied by photovoltaic panels that have been sized according to the compression chiller’s nominal capacity (photovoltaic nominal capacity assumed to be the chiller’s nominal capacity divided by its coefficient of performance, its nominal electric consumption). In another case, an absorption chiller has been utilized. Solar Fresnel panels are designed to theoretically supply the energy required for the conventional single-stage absorption chiller to operate. Additionally, a natural gas boiler is included to ensure the proper inlet temperature for the chiller’s heat input. This article is an extension of the previous works of authors [14].

The objective of this paper is to compare the behavior of these two technologies in different scenarios:

• Under different demand profiles. Taking a variable and a constant approach.
• Under different sizing considerations. Measuring the impact of photovoltaic panels and in the case of absorption chillers, how much impact has the possibility of having a usable surplus of energy to supply the heating needs of the equipment.

To address these subjects, the article will start in Section 2 by defining the technologies considered, which combinations of systems will be simulated, the location assumed for the study, and how the demand profiles were generated for the different cases. A brief explanation of the simulation methodology will be provided in Section 3 to understand which results were calculated and how they were obtained. Finally, results will be analyzed and discussed thoroughly in the results section (Section 4).

Different simulation approaches have been employed to determine the optimal plant sizes, enabling a technical and economic comparison of the results. Simulation models have been developed for both configurations using the modular methodology established in the WEDISTRICT project [15]. Based on the results obtained from these simulations various key parameters—technical, environmental, and economic—have been calculated and analyzed [16].

The main novelty of this article is the comparison between a solar-driven absorption chiller and a photovoltaic compression chiller under two demand profiles: variable and constant. The variable demand profile is representative of a residential setting, whereas the constant demand profile is typical of data centers. This article aims to provide a clear view of the impact of demand profile and situational components in the selection between absorption and compression chillers.

2. System Description

2.1. Technologies Considered

This paper aims to compare two different options for cooling generation. On one side, a less commonly used system utilizes an absorption chiller, which needs a heat input at a temperature of around 95 °C to go through its absorption cycle (see Figure 1). This absorption cycle uses a mixture of LiBr with water, which varies its concentration along the absorption cycle working at partial vacuum pressures to be able to evaporate water at the needed temperatures for cooling. Inside the absorption chiller simulation, a buffer tank is assumed to be present to keep a more stable operation.

The other configuration being studied is a compression chiller paired with photovoltaic panels to deal with its electrical consumption (see Figure 2). There is no battery storage system, and the benefits of the commercialization of the surplus electricity generated will not be considered.

One of the main aspects to consider when comparing these systems is the coefficient of performance (COP). Absorption chillers have lower COP values due to their theoretical limitations. Their main advantage lies in their energy consumption, which is primarily thermal. Thermal energy usually costs less than electrical energy. In contrast, compression chillers use electrical energy for their operation. For this analysis, a single-stage absorption chiller has been used, reaching an assumed COP value of 0.75. While for the compressor chiller, a value of 3 was assumed [17].

Gas boiler efficiency has been considered constant, with a value of 95% at all operation points. Storage tank losses are fixed to an overall thermal loss coefficient value of 0.16 ${\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2·\mathrm{K}}}$, according to literature [18].

Photovoltaic panels are considered to have 2.58 m² each, a 550 W peak power (technology currently being installed), with an efficiency of 21.3% in line with a competitive company data sheet [19]. These panels have been set with a 40-degree slope (equal to latitude) to optimize production [20]. Finally, thermal solar panels are considered to be Soltigua’s Fresnel collectors (Soltigua^TM, Gambettola, Italy) which have an intercept efficiency of 67%, a first-order loss coefficient of 0.032 ${\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2·\mathrm{K}}}$, and a second-order loss coefficient of 0.00018 ${\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2·{\mathrm{K}}^2}}$. These collectors have an area of 148.5 m² each [21].

2.2. Cases Evaluated

Both alternatives have been fixed in their cooling capacity to meet the demand proposed, to make the comparison fair. Both nominal capacities have been set to 500 kW. A parametric study for the renewable heating generation technologies for the absorption chiller has been carried out to select the proper sizing. This study altered only solar generation as the boiler was left at the nominal capacity needed to avoid idle hours for the absorption chiller (667 kW). The photovoltaic (PV) panels capacity was fixed to the compression chiller’s capacity divided by its COP (its nominal electric consumption, 166.67 kW).

Firstly, Levelized Cost of Energy (LCOE) and CO₂ emissions have been defined in the context of the absorption chiller. Special attention has been given to the amount of power required as heat input for the system and how variations in the supply can affect it. Then, efforts have been made to establish the preferred technology among users. The LCOE and CO₂ emissions for the compression chiller have been calculated, and studies have been conducted to determine the improvements the system can achieve by using photovoltaic (PV) solar panels to provide electricity.

The same analysis was then repeated for the same cases but considering the constant demand profile mentioned. This constant demand is considered to be of special interest as data centers show these profiles in their yearly operation [22]. The different cases are summarized in

Table 1. Cases evaluated.

Cooling Technology	Demand	Parametric Study	PV Panels	Heat Source Price Considered
Absorption chiller	Variable	Yes	-	Yes
Absorption chiller	Variable	No	-	No
Absorption chiller	Constant	Yes	-	Yes
Absorption chiller	Constant	No	-	No
Compression chiller	Variable	No	Yes	-
Compression chiller	Variable	No	No	-
Compression chiller	Constant	No	Yes	-
Compression chiller	Constant	No	No	-

.

2.3. Location and Demand

This study was conducted in Madrid, where there is a significant solar resource available (as shown in Figure 3), and there is a seasonal cooling demand during summer.

Two demand profiles have been tested: one with variable demand, based on an annual consumption of approximately 751 MWh/year, which corresponds to cooling peaks of 500 kW; and a profile that considers a constant demand of 500 kW. This constant value will have a special interest when considering year-long consumers such as data centers.

The variable cooling profile is derived from TMY (typical meteorological year) for Madrid in this case. The TMY data used in this instance have been obtained from the Joint Research Centre Photovoltaic Geographical Information System (PVGIS), from the TMY tab (period 2005–2020), offering hourly ambient temperature and other meteorological parameter values [23].

From this data and having the following assumptions, a cooling demand profile is generated.

1

For Madrid, a “cooling season” (warm months where cooling demand peaks and heating demand tends to be domestic hot water production alone) is defined. This has been chosen by looking at the temperature profile from the climate data obtained from Figure 4. And assuming that cooling production occurs between dates that the daily average ambient temperature of the three last consecutive days remains above 15 °C and is turned off when this condition is not fulfilled two days in a row (4 May to 29 September).

2

Within the cooling season, an occupation schedule is assumed, indicating when the building is occupied. Indeed, two occupation schedules are used, depending on the time of the year:

• Occupation schedule 1: A high cooling season, corresponding to the warmest months in summer. For this period, a value of 1 for the weight factor was used for the occupied hours (from 6 a.m. to 6 p.m. on weekdays and 8 a.m. to 4 p.m. on weekends).
• Occupation schedule 2: Shoulder cooling periods, corresponding to the late spring/early autumn months. For this period, a value of 0.8 for the weight factor was used for the occupied hours (same assumed occupation hours as occupation schedule 1).

3

The dimensionless hourly cooling load is estimated based on conversion tables (one for hours with occupation, the other for hours with no occupation) relating the ambient temperature with the fraction of peak cooling load, as shown in Figure 5. The peak cooling load (100%) is assigned to the highest hourly ambient temperature. Hourly dimensionless cooling demand profile is developed for Madrid following Equation (1).

$$CD{P}_{Dimensionless}\sim Sc{h}_{weighting.f}·Pea{k}_{percentage}$$

(1)

where:

• $CD{P}_{Dimensionless}$: Dimensionless cooling demand profile;
• $Sc{h}_{weighting.f}$: 1 or 0.8 according to described schedules;
• $Pea{k}_{percentage}$: Taken from Figure 5.

Figure 4. Hourly and daily average ambient temperature based on TMY for Madrid [23].

Figure 4. Hourly and daily average ambient temperature based on TMY for Madrid [23].

Figure 5. Percentage of peak cooling load as a function of the ambient temperature.

Figure 5. Percentage of peak cooling load as a function of the ambient temperature.

The hourly demand distribution, which is determined by the methodology previously explained, and the descending monotonic curve are represented in Figure 6 and Figure 7. These figures enable the necessary capacity to be determined.

Figure 6 shows how cooling demand starts appearing from early months of the year (around 14 April) with low demand values, generating what is usually called “shoulder months”. This period shows demand values around 200 and 350 kW which usually makes equipment designed for peaks work in partial load ratios, or start intermittently. From hour 4000 (around 16 June) up to hour 6500 (end of 28 September) a higher demand is visible. The rest of the cooling period shows once again a “shoulder month” behavior. It is visible that these higher loads correspond to the summer months (higher ambient temperatures, Figure 4) but also with the highest average solar irradiation months (Figure 3).

3. Methodology

3.1. Software

TRNSYS (TRaNsient SYstems Simulation Program) software [24] was utilized following the modular methodology specifically developed for the WEDISTRICT project [25]. This methodology employs TRNSYS macros and decks. Macros consist of a series of TRNSYS types that are used in combination to minimize the number of necessary connections.

The macros used by this methodology have the following features that improve modularity and flexibility:

1

Nomenclature: Guidelines for naming macros, types, and variables are established.

2

Variables used for input and output are exchanged between different macros via equation blocks. This method streamlines the process, minimizes the number of connections, and facilitates the efficient replacement of one macro with another, requiring changes to only a few connectors.

3

Parametrization procedure: A Python script has been developed to update the input data used in each simulation.

4

Control strategy: Each macro features a specific control strategy based on its technology. This approach allows the same macro to be integrated into different systems, reducing the number of control parameters that need to be configured.

5

Results: Each macro presents its own results along with internal calculations, including energy and mass balance.

The compression chiller macro (M4300) is presented as an example. This macro supports two distinct operation modes: one in which the chiller operates in series to achieve a specified output temperature and another where the chiller works in parallel with a cold water storage tank to maintain a setpoint within the tank. The representation in TRNSYS of the compression chiller macro is shown in Figure 8. The blue arrows represent the heat dissipated by the chiller (QCHI01) and the pipe losses (QLsPI01). The purple arrows indicate the power consumed by the electrical chiller (WCHI01), and the pump (WPU01). The mass flows are represented by black arrows, with MIn01/TIn01 showing the incoming mass flow and temperature, and MOu01/TOu01 displaying the outgoing mass flow and temperature.

This methodology is used to develop macros that facilitate the faster generation of decks to be executed by TRNSYS. After running the simulations, all results are collected, and Key Performance Indicators (KPIs) are calculated using a Python 3.7 script.

TRNSYS Simulation Models

TRNSYS components are chosen from the TRNSYS libraries as well as from the “Thermal Energy Systems Specialists” library (TESS) and improved with manufacturers’ performance data or supplied information.

The absorption macro simulation model uses “Type 107” from TESS libraries as the central type. This type considers an external performance map, which varies the absorption chiller’s partial load ratio considering source temperatures and set temperatures. This performance map is taken from the standard information from the type, which takes into account hot water being used as the heating source.

The generation side of this macro was supplied by Fresnel solar collectors and gas boilers. Concentrating collector macro used, takes “Type 1245”, from TESS libraries as well, as the central type. This type takes into account solar irradiation, in this case, filtered by two different shading masks, one by horizon shading and the other one by the own collector array shading types (TRNSYS “Type 67” and TESS “Type 1262”), automated by this macro parametrization. “Type 1245” uses data from Soltigua’s Linear Fresnel collectors [21] through an external file which provides Incidence Angle Modifiers (IAMs) mapping. Also, parametrization values for optical efficiency are set automatically in the macro (see Equations (2) and (3)) (as mentioned in Section 2.1). This macro is also able to account for Soltigua’s parabolic trough collectors (PTC) through the use of a technology selection parameter.

$$\eta =a_0-a_1·{\displaystyle \frac{(\Delta T)}{I_T}}-a_2·{\displaystyle \frac{{(\Delta T)}^2}{I_T}}$$

(2)

$${\eta }_{Fres,Solt}=0.67-0.032·{\displaystyle \frac{(\Delta T)}{I_T}}-0.00018·{\displaystyle \frac{{(\Delta T)}^2}{I_T}}$$

(3)

where

• $a_0$: Intercept (maximum) collector’s efficiency [-];
• $a_1$: Negative of the first-order coefficient in collector efficiency equation [${\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2·\mathrm{K}}}$];
• $a_2$: Negative of the second-order coefficient in collector efficiency equation [${\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2·{\mathrm{K}}^2}}$];
• $\Delta T$: Temperature difference between the collector’s inlet temperature ($T_i$) and the ambient temperature ($T_{amb}$) [K];
• $I_T$: Effective total radiation incident on the solar collector tilted surface per unit area including the impact of off-normal solar radiation (incidence angle modifiers) [${\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2}}$].

For the gas boiler TESS “Type 751” was used. This type also takes an external mapping file for boiler part load ratio (PLR) efficiency. In the particular case of the boiler macro, different fuels can be considered through parametrization. Natural gas was selected for this case. For the compression chiller simulation TESS “Type 655” was used, which also has two external files, one to account for different operation conditions and the other to account for different PLRs. For this paper, Climaveneta’s chiller information for files was considered [17]. The Type 103 is used to simulate the PV panels.

Thermal energy storage for the Fresnel collector and the internal absorption chiller buffer utilizes TESS “Type 534” to establish water as the storage medium. The calculated volume is parameterized based on the capacity of the equipment being used, while also taking into account the lateral, bottom, and edge loss coefficients.

Most temperature controls take TRNSYS “Type 2” (on/off differential controller) as a way of comparing variable temperatures with dead bands to provide stability for the simulation. Flows are set with TRNSYS “Type 110”, known as the variable speed pump to set pump flows and establish an electrical power demand for pumps in different macros. Lastly, some macros have internal piping to consider equipment inertia, these pipes are simulated using TESS “Type 709” which simulates the thermal behavior of fluid flow in a pipe or duct using variable-size segments of fluid.

3.2. KPIs Definition

All KPIs have been calculated according using the equations provide by reference [16]. The comparison is made based on the Levelized Cost of Energy (LCOE) and CO₂ emissions.

3.2.1. Levelized Cost of Energy

The Levelized Cost of Energy (LCOE) evaluates the average net present value of energy expenditures over the system. It is especially useful for comparing different alternatives, in cases where significant initial investments are needed but operating costs decrease over time. This scenario often occurs in systems that rely heavily on renewable energy sources. The LCOE allows to determine the cost associated with per unit energy generation. Essentially, the LCOE indicates the average price that consumers would need to pay to cover all associated costs while achieving a rate of return corresponding to the chosen discount rate.

The LCOE is expressed by Equation (4) [16]:

$$LCOE={\displaystyle \frac{CAPEX·CRF+OPE{X}_f+OPE{X}_v}{Q_C}}$$

(4)

$$CRF={\displaystyle \frac{i{(1+i)}^n}{\left[{(1+i)}^n\right]-1}}$$

(5)

• LCOE: Levelized cost of cooling energy [EUR/MWh];
• CAPEX: Capital expenditure for the equipment [EUR/MWh];
• CRF: Capital recovery factor;
• OPEX_f: Fix operational costs for cooling [EUR/year];
• OPEX_v: Variable operational costs for cooling [EUR/year];
• Q_C: Cooling energy supplied per year [MWh/year];
• i: interest rate;
• n: project lifetime and number of annuities received.

Table 2. Parameters used to calculate the level of energy cost.

Case	Value	Reference
Specific cost for TES [EUR/m3]	235	[28]
Specific cost for compression chiller [EUR/kW]	196	[29]
Specific cost for PV [EUR/kW]	1100	[30]
Specific cost for natural gas boiler [EUR/kW]	60	[31]
Specific cost for Fresnel collector [EUR/m2]	200	[32]
Specific cost for absorption chiller [EUR/kW]	472	[33]
OM fixed [%]	5	[7]
Discount rate [%]	8.5	[34]
Life [year]	25	[34]
Price natural gas [EUR/MWh]	54.1	[26]
Price electricity [EUR/MWh]	142.6	[27]

presents the economic input data considered to calculate the LCOE. The electricity and gas prices are taken from the EU statistical website [26,27]. These values correspond to the price of electricity gas in Spain for the second half of 2024. The specific costs of the thermal energy storage are an average between the range indicated in reference [28]. Fixed operating and maintenance costs are calculated as a percentage of the investment costs.

3.2.2. CO₂ Emissions Coefficient

This CO₂ emissions coefficient represents the emissions per unit of energy generated and is calculated using the Equation (6)

$$k_{{CO}_2}={\displaystyle \frac{{\sum }_i{E}_i·k_i}{Q_C}}$$

(6)

where

• $k_{{CO}_2}$: CO₂ emission coefficient [${\mathrm{kg}}_{{\mathrm{CO}}_2}$/MWh];
• $E_i$: energy supplied by energy carrier i per year [MWh/year];
• $k_i$: Emissions coefficient of energy carrier i [${\mathrm{kg}}_{{\mathrm{CO}}_2}$/MWh];
• $Q_C$: Cooling energy supplied per year [MWh/year].

For CO₂ emissions,

Table 3. CO₂ emissions factors considered [35,36].

Energy Carrier	CO2 Emissions Factor [kg CO2/MWh]
Fossil Fuel	220
PV electricity	0
Solar thermal	0
Grid electricity	165

has been taken as seen in the reference [16] from [35]. With the exception of the highlighted value for electricity of 420 kg CO₂/MWh, that was corrected to 165 kg CO₂/MWh considering the value shared by the European Environment Agency for Spain for 2021 [36].

4. Results

4.1. Absorption Chiller

First, the parametric study to determine the optimal solar field and the associated thermal energy storage has been made, the study was performed using the variations described in

Table 4. Parametric simulation of solar field and storage.

Technology	Initial Value	Final Value	Step	Simulations
Solar field [m2]	1500	3000	250	7
Storage volume [m3]	100	1000	25	36

.

The distribution of the results for the variable is shown in Figure 9. From the results obtained the selection was aimed at achieving the least amount of emissions possible. As visible in Figure 9, the results cluster in sets along the plot. These sets are fixed by the increasing sizes of solar panels, as the size increases, LCOE values tend to grow due to the CAPEX of the additional solar collector area. Inside these clusters of cases, as storage volume is set, smaller collectors show little advantage from having bigger tanks (emissions results show better results in smaller tanks). But when solar sizes start having a big impact on the energy provided to the absorption chiller (around 2000 m²), a variation in tank sizes starts having a bigger spread in CO₂ emissions, proving that a higher emission boiler operation is being displaced by collectors operating with bigger tanks. From an environmental point of view, this behavior shows an optimal value for a solar field of 3000 m² and a 550 m³ storage (selected case), which shows that while incrementing the number of square meters dedicated to the solar field at this point, solar generation would reach a point when it starts saturating its production, and due to its size starts dumping energy instead of using it, increasing the electric consumption of the system without reducing the boiler operation, thus incrementing CO₂ emissions for the system as it is visible for cases with LCOEs higher than the design selected. Other considerations could be made when selecting a design case, smaller solar fields could be selected in order to prioritize economic results.

The optimal results for both demand profiles can be seen in

Table 5. Absorption chiller optimum cases for variable demand profile considering heat source costs.

Solar Field [m2]	TES Volume [m3]	Emissions Coef. [kg CO2/MWh]	LCOE [EUR/MWh]
1500	100	222.56	211.06
1750	150	210.87	222.25
2000	225	202.25	235.34
2250	275	195.67	247.83
2500	425	189.87	265.02
2750	525	185.11	280.14
3000	550	181.08	291.87

and

Table 6. Absorption chiller optimum cases for constant demand profile considering heat source costs.

Solar Field [m2]	Tank Volume [m3]	Emissions Coef. [kg CO2/MWh]	LCOE [EUR/MWh]
1500	100	284.97	103.42
1750	125	278.48	103.89
2000	200	272.03	104.77
2250	250	265.55	105.44
2500	300	259.07	106.12
2750	400	252.56	107.15
3000	450	246.06	107.60

. It can be observed a relationship between economic and environmental optima in two extreme cases. The case with the lowest LCOE also has the highest emissions factor, while the case with the lowest emissions factor corresponds to the highest LCOE. For this study, the cases with the lowest emissions factor are selected.

Sometimes it is possible to find industrial facilities where excess energy is constantly being wasted, so it would be interesting to recover it. So, it is interesting to examine the scenario in which a free, zero-emission heat source supplies thermal energy to the absorber. In this case, the solar field, along with its storage system and boiler, is excluded from the diagram shown in Figure 1. Under this assumption, a Levelized Cost of Energy (LCOE) of 115.47 EUR/MWh and an emission factor of 50.41 kg CO₂/MWh are achieved for the case of variable demand, while for the case of constant demand, an LCOE of 29.64 EUR/MWh and an emission factor of 20.05 kg CO₂/MWh are obtained.

Table 7. Results for the absorption chiller cases.

	Var. Profile with Heat	Cons. Profile with Heat	Var. Profile Without Heat	Cons. Profile Without Heat
CAPEX [EUR/MWh]	130.78	21.77	47.52	8.95
OPEX [EUR/MWh]	161.09	85.83	67.95	21.49
LCOE [EUR/MWh]	291.87	107.60	115.47	29.64
$k_{{CO}_2}$ [kgCO2/MWh]	181.08	246.06	50.41	20.05

presents the cases selected for both demand profiles, considering the two proposed scenarios: heat supplied by the Fresenal collector with TES and natural boiler, and heat provided by a free, zero-emissions resource. Comparing the two demand profiles reveals that the variable demand case has a lower emission factor of 26.41%, while the constant demand profile case exhibits a 63.13% lower LCOE. The higher emissions factor for the constant demand profile. The high emission factor associated with constant demand arises from the low solar field production during the winter months. As a result, the natural gas boiler provides a large portion of the thermal energy required by the absorption chiller. In contrast, the reduction in the Levelized Cost of Energy (LCOE) for the constant demand scenario occurs because, with the same equipment and similar investment costs, there is a substantially higher cooling demand generated. As can be observed the variable demand scenario will be more sensitive to changes affecting CAPEX, while the constant demand scenario will be more sensitive to changes affecting OPEX. It is evident that the absorption chiller has a more favorable cost and a lower CO₂ emissions coefficient when the energy resource is free and has zero emissions. The solar panels and a natural gas boiler need to be installed to provide the heat required for the absorption chiller, which increases the levelized cost of electricity (LCOE) due to the high costs associated with both purchasing (CAPEX) and operating (OPEX) this equipment. Additionally, it raises the CO₂ emissions associated with producing the needed heat. These CO₂ emissions can be observed to originate from the use of natural gas as a heat source, having the biggest impact on this indicator.

4.2. Compression Chiller

Table 8. Results for the compression chiller cases.

	Var. Profile with PV	Cons. Profile with PV	Var. Profile Without PV	Cons. Profile Without PV
CAPEX [EUR/MWh]	32.14	5.51	11.2	1.92
OPEX [EUR/MWh]	48.93	64.86	86.94	79.16
LCOE [EUR/MWh]	81.07	70.37	98.13	81.08
$k_{{CO}_2}$ [kgCO2/MWh]	35.13	46.57	62.42	56.84

shows the results for the compression chiller. For each deamand profile, two scenarios are analyzed: with PV and without PV. In cases with PV, a comparison of the two demand profiles shows that the variable demand case has a lower emission factor, while the constant demand profile case has a lower Levelized Cost of Energy (LCOE). The higher emission factor in the constant demand case occurs because a larger proportion of the electricity consumed is sourced from the grid. In contrast, the reduction in LCOE for the constant demand case is due to the fact that a greater amount of cooling is produced using the same equipment. These results demonstrate that the installation of solar panels improves the Levelized Cost of Energy (LCOE) values, with a reduction of 17.38% for the variable demand profile and 13.20% for the constant demand profile. Additionally, this investment reduces CO₂ emissions by 43.72% for variable demand profile and by 18.06% for the constant demand profile. This improvement can be attributed to the benefits gained from electricity generated by the solar panels throughout the year. As can be observed, the price of electricity will have a significant impact on all scenarios. If the price increases, the LCOE will also rise, while a decrease in price will lead to a reduction in the LCOE. Changes in CAPEX will primarily influence variable demand with PV; in other cases, the LCOE will be less affected.

4.3. Discussion

Figure 10 compares the results obtained for each cooling system under the variable demand profile. The case with the lowest LCOE and emission factor is the compression chiller with PV panels. In this case, the PV panels supply the majority of the demand for the compression chiller. According to the results shown in Table 7 and Table 8, the capital costs of the compression chiller cases are lower than those of the absorption chiller cases.

Figure 11 compares the results obtained for each cooling system under the variable constant profile. In this demnad profile, the lowest LCOE and emission factor are achieved thanks to the absorption chiller powered by a free, emission-free energy source. The low LCOE is primarily because the only significant cost is the capital expense related to the absorption chiller. The variable operating and maintenance costs are minimal and arise from the low electricity consumption of auxiliary equipment, such as circulation pumps and cooling towers. This low electricity usage also contributes to the low emission factor.

5. Conclusions

The performance of absorption and compression chillers has been analyzed in four scenarios for each type. For absorption chillers, the use of existing thermal sources under variable and constant demand profiles has been analyzed. In contrast, the compression chiller analysis focused on the impact of integrating photovoltaic (PV) panels, also under both demand profiles. The results suggest that absorption chillers are the best alternative for constant demand situations when a nearby heat source is available. Their generator outlet typically reaches around 85 °C, making it a viable energy source and a promising area for further research, particularly for data centers. For variable demand scenarios, the compression chiller is the better option, especially when PV panel installation is feasible.

The findings of this study are limited to the specific location and the economic values that were assumed. To draw more reliable conclusions, it would be beneficial to expand this study to include additional locations. Additionally, conducting a sensitivity analysis of the financial parameters could help identify potential areas for improvement. Another area for future work is the addition of a battery for electricity storage, along with a cold energy storage system. By integrating these two components, the technical, economic, and environmental performance of the proposed system will be enhanced, optimizing the capacity of each piece of equipment.