Development of a Dual Fuel ICE-Based Micro-CHP System and Experimental Evaluation of Its Performance at Light Loads Using Natural Gas as Primary Fuel

Abstract

This study presents the implementation of a micro-generation system and its operation procedure, based on a dual fuel diesel engine using natural gas as the primary fuel and conventional diesel as the pilot fuel. On the other hand, the evaluation and validation results by experimental testing of a model according to International Energy Agency—IEA—Annex 42, applied to dual fuel diesel micro-cogeneration system, are also presented. The control procedure for experimental operation depends of both inputs’ electric power generation demand and desired substitution level due a given natural gas availability. The heat recovery system of the micro-generation system uses a gas–liquid compact heat exchanger that was selected and implemented, where wasted heat from exhaust gases was transferred to liquid water as a cool fluid. Effective operation engine performance was determined by measurement of masses’ flow rate such as inlet air, diesel and natural gas, and also operation parameters such as electric power generation and recovered thermal power were measured. Electric power was generated by using an electric generator and then dissipated as heat by using an electric resistors bank with a dissipation capacity of $18\mathrm{kW}$. Natural gas fuel was supplied and measured by using a sonic nozzle flowmeter; in addition, natural gas composition was close to $84.7\%$ ${\mathrm{CH}}_4$, $0.74\%$ ${\mathrm{CO}}_2$ and $1.58\%$ ${\mathrm{N}}_2$, with the rest of them as higher hydrocarbons. The highest overall efficiency (electric efficiency plus heat recovery efficiency) was $52.3\%$ at $14.4$ kW of electric power generation and $0\%$ of substitution level. Several substitution levels were tested at each engine electric power generation, obtaining the maximum substitution level of $61\%$ at $17.7$ kW of electric power generation. Finally, model prediction results were closed to experimental results, both stationary and transient. The maximum error presented was close to $20\%$ associated to thermal efficiency. However, errors for all other variables were lower than $10\%$ for most of micro-cogeneration system operation points.

1. Introduction

Compression ignition (CI) engines operating in dual-fuel mode are a reliable technological strategy for power generation, reduction of heavy fossil fuel consumption, and attenuation of carbon dioxide and soot emissions. The so-called “dual fuel engines” have been extensively tested around the world to quantify the mentioned benefits when compared with typical diesel fuel operation, mostly at constant load frameworks, leading to a well-established international knowledge on performance and emission characteristics of dual fuel engines for power generation. Dual fuel mode means that two fuels are used for heat release during combustion, and although many options could be considered for this purpose, usually a gaseous fuel named “main fuel” [1] provides around $60\%$ to $97\%$ of the heat released and diesel acts as “pilot fuel” or ignition source for the air-gaseous fuel mixture and provides the rest of the heating energy [2]. Diesel substitution is one of the most important parameters to evaluate the dual fuel performance, and a bigger share of gaseous fuel promotes premixed combustion in the air gaseous fuel mixture, leading to reductions in soot emissions [3,4]. Depending on how the main fuel is admitted, dual fuel engines are classified as “port-fuel injected” if this fuel is admitted along with air in the intake system, and “direct-fuel injected” if the main fuel is admitted into the cylinder when the piston is close to the top dead-center position [5]. In dual fuel engines, the intake system is usually designed to admit a premixed air gaseous fuel mixture; however, direct injection of the gaseous fuel during the compression stroke has been used in some research projects and commercial engines [6,7].

Natural gas as gaseous fuel is used in many applications such as furnaces, water heaters, cooking stoves, processing foods, air conditioning, generating steam in water boilers then used in steam turbines, power generation in gas turbines and internal combustion engines both stationary and vehicular [8]. The composition of natural gas depends generally on its source of extraction, thus its heating value usually varies from $26.000$ kJ/kg (700 Btu/scf) to $59.000$ kJ/kg ($1.600$ Btu/scf) [9]. On the other side, natural gas is mostly composed of >85% methane (${\mathrm{CH}}_4$), 3–8% ethane (${\mathrm{C}}_2{\mathrm{H}}_6$), <1% propane (${\mathrm{C}}_3{\mathrm{H}}_8$), <2% heaver hydrocarbons as butane (${\mathrm{C}}_4{\mathrm{H}}_{10}$) and pentane (${\mathrm{C}}_5{\mathrm{H}}_{12}$), 1–2% carbon dioxide (${\mathrm{CO}}_2$), 1–5% nitrogen (${\mathrm{N}}_2$) and with minor constituents such as helium (${\mathrm{H}}_2$) and hydrogen sulfide (${\mathrm{H}}_2\mathrm{S}$) [8,10]. Natural gas is considered a cleaner fuel compared with other heavier fossil fuels, thereby offering important environmental benefits because of a significantly lower production of particulate matter ($\mathrm{PM}$), oxides of nitrogen ($\mathrm{NOx}$) and hydrocarbons ($\mathrm{HC}$) than conventional diesel and gasoline combustion engines [11,12,13].

Aiming to enhance the dual fuel diesel engine performance keeping high substitution levels, some strategies to improve the auxiliary components of the engine may be implemented such as a direct common-rail injection system, injection delay setting depending on load and speed, and even turbochargers [14]. However, dual fuel diesel engines could be seen as double power generation source systems, combining electric or mechanical power generation with thermal power generation, where this thermal power can be recovered from the water coolant and exhaust gases, allowing them to replace the installation of electric heaters and gas burner plants when combined application is needed [15,16]. Some processes can need heating, such as textile products drying, dehydration of vegetables or fruits, water and air heating or substances. The engine exhaust gases can achieve temperatures close to 150∼450 °C depending on the load and substitution level in dual fuel engines, having a high sensible heat recovery potential [17,18]. Thereby, the strategy to generate combined electric and thermal power from recovering the exhaust gases or water coolant heat can lead to increase the thermodynamic efficiency of the system, bringing a lower overall fuel consumption, where combined production of heat and electric power refers commonly to cogeneration [19].

Cogeneration usually refers to systems able to produce simultaneously both mechanical or electric energy and useful thermal energy from the same primary energy source, where the thermal energy is recovered from wasted heat from the combustion process of engines or turbines, even in some cases from high temperature fuel cells [20,21]. If thermal energy demand is taken into consideration, installed on-site electric power generation has an important thermodynamic efficiency advantage compared to centralized power generation, as wasted heat energy from the electric power generation process can be used [22]. Often, the centralized power generation plants do not use these wasted heat energies and these same energies must be ejected into the environment, representing economic loss coupled with greater specific emissions releasing. When an on-site power generation plant recovers wasted heat energy for using on the process itself, then the thermodynamic efficiency rises, becoming a fuel-economic advantage, and this thermodynamic efficiency is also named overall thermal efficiency, which relates electric power and thermal power to chemical fuel energy based on lower heating values [19,22].

In cogeneration, the overall thermal efficiency of energy conversion increases depending on the prime mover cycle, and internal combustion engines and Stirling engines have an electric efficiency close to 20–35% but obtain an overall thermal efficiency over 45% up to 80%, micro steam turbines have an electric efficiency close to 10–20% but obtaining an overall thermal efficiency over 65% up to 80%, micro gas turbines have an electric efficiency close to 15–30% but obtain an overall thermal efficiency over 60% up to 80% [23], and finally, fuel cells have an electric efficiency close to 30–70% or 25–40% but obtain an overall thermal efficiency over 60% up to 80% depending on the technology of the fuel cell, SOFC (Solid Oxide Fuel Cell) or PEMFC (Polymer Electrolyte Membrane Fuel Cell), respectively [24].

Models have been developed for applying and control of cogeneration systems depending on operation parameters such as electric efficiency, thermal efficiency associated to heat recovered, mass flow of heat recovery fluid, temperatures of heat recovery fluid, properties and mass flow rate of fuel and working fluid, and thermal and geometric parameters of both the heat recovery device and electric power generation system [25]. These models have to be as simple as possible, but also able to predict the real operation of the micro-cogeneration system. These models can be experimentally calibrated to try to obtain the operation maps and variables prediction as described above [26,27].

Therefore, this study arises as an experimental validation of a previous study, which develops and calibrates a dual fuel engine micro-cogeneration (DECOG) model having as the main input parameters the electric power generation and substitution level. The present study shows the implementation and experimental evaluation of a micro-cogeneration system and also its model validation for a dual fuel diesel-engine-based micro-cogeneration system (electric power close to 15 kW) using natural gas as the primary fuel and diesel as the pilot fuel, where the thermal power is recovered only from wasted heat from exhaust gases. The applied models for this study are dependent on electric power generation and substitution level as the main operating parameters in a dual fuel diesel engine. Experimental operation maps allow us to calibrate the models for subsequent evaluation and validation. Additionally, a transient validation and prediction capacity is presented based on the outlet heat recovery fluid temperature as the main control parameter for micro-cogeneration system implementation in real applications. Additionally, a description of the operation control procedure is explained, which depends on both inputs of electric power generation demand and substitution level for a given natural gas availability, then sensor and actuators are set electronically by a computer achieving as closed as possible the desired inputs and measuring output variables as recovered heat.

2. Models Theoretical Description

A model for an internal combustion engines-based cogeneration system has been developed initially within International Energy Agency (IEA) Annex 42 for usages applied to building simulation programs, but then modified and adapted by DECOG for a dual fuel diesel-engine-based micro-cogeneration system [28]. The model aims to accurately predict the electric and thermal power outputs of any internal-combustion-engines-based cogeneration system. The correct operation of the model requires a specified number of coefficients found experimentally, so that this model must be calibrated and validated by experimental results, which will have applicability only to the implemented engine type, implemented heat recovery devices and fuel for the testing operation [29]. The model defines the control volume for the cogeneration device as Figure 1 shows and it is explained below [25].

• Energy conversion control volume (ECCV) involves the engine working fluid, exhaust gases and engine alternator, providing information from the combustion engine unit operation map to the model.
• Combustion engine control volume (CECV) involves the aggregated thermal capacitance associated with the engine block and heat recovery device shells.
• Heat recovery water control volume (HRWCV) involves the recovery water flowing through the devices and heat recovery devices in thermal contact.

2.1. Dual Fuel Diesel Engines

In dual fuel diesel engines operation, the gaseous fuel as the primary fuel enters together with intake air; however, it does not auto-ignite by itself. Ignition of that premixed gas is achieved when a small diesel fuel quantity is injected as a pilot fuel, timed at the end of the compression stroke [18]. The pilot fuel auto-ignites first, inducing mixture with the primary fuel, and combustion air ignites as well. Later on, gaseous fuel combustion occurs via a flame initiation at an unspecified location in the combustion chamber [30,31]. Thus, depending on the primary fuel quantity, a pilot fuel quantity must to be injected, which is why a substitution level Z is defined for dual fuel operation as the percentage between the quantity of diesel fuel reduction during dual mode and conventional diesel fuel quantity at a given power [5,32], as shown in Equation (1).

$$Z={\displaystyle \frac{{\dot{m}}_{D,c}-{\dot{m}}_D}{{\dot{m}}_{D,c}}}·100$$

(1)

where ${\dot{m}}_{D,c}$ is the diesel mass flow rate in conventional operation, i.e., at $0\%$ substitution level, ${\dot{m}}_D$ is the diesel mass flow rate in dual fuel operation mode. Due to dual fuel operation, equivalence ratio $\varphi$ can be redefined as a function of both fuels, natural gas and diesel. Equation (2) shows the equivalence ratio for dual fuel operation mode [30].

$$\varphi ={\displaystyle \frac{{\displaystyle \frac{{\dot{m}}_D}{{\varphi }_{D,stq}}}+{\displaystyle \frac{{\dot{m}}_{NG}}{{\varphi }_{NG,stq}}}}{{\dot{m}}_a}}$$

(2)

where ${\varphi }_{D,stq}$ and ${\varphi }_{NG,stq}$ are the stoichiometric fuel–air ratio for diesel and NG, respectively, and ${\dot{m}}_{NG}$ and ${\dot{m}}_a$ are the NG and combustion air mass flow rates, respectively.

2.2. Energy Conversion Control Volume

A steady state energy balance can be applied to the energy conversion control volume to obtain Equation (3). Such equation relates the air and fuel total enthalpy as equal to the sum between electric power generation, heat power production and exhaust gases total enthalpy.

$${\dot{H}}_a+{\dot{H}}_f=P_E+P_T+{\dot{H}}_{exh}$$

(3)

where ${\dot{H}}_a$, ${\dot{H}}_f$ and ${\dot{H}}_{exh}$ are the air, fuel and exhaust gases total enthalpies, respectively, and $P_E$ and $P_T$ are the electric power generation and thermal power recovered, respectively. However, the model does not aim to complete characterization of the energy balance described by Equation (3). Instead, the engine performance is related to total chemical power $P_Q$ added by both gaseous and pilot fuels, so that natural gas and diesel are included as shown in Equations (4) and (5).

$$P_E={\eta }_E{P}_Q={\eta }_E\left({\dot{m}}_{D}LH{V}_D+{\dot{m}}_{NG}LH{V}_{NG}\right)$$

(4)

$$P_T={\eta }_T{P}_Q={\eta }_T\left({\dot{m}}_{D}LH{V}_D+{\dot{m}}_{NG}LH{V}_{NG}\right)$$

(5)

where ${\eta }_E$ and ${\eta }_T$ are the electric and thermal efficiencies, $LH{V}_i$ is the lower heating value of fuel species i, ${\dot{m}}_i$ is the mass flow rate of fuel i, and subscripts D and $NG$ represent the diesel and natural gas, respectively. In steady state operation, the electric and thermal efficiencies can be determined by using empirical correlations, which could relate variables as electric power, heat recovery water flow rate and temperature of the same heat recovery water [29]. Nevertheless, electric and thermal efficiency correlations can be redefined as a function of electric power and substitution level, since the current study applies a dual fuel diesel engine for the micro-cogeneration system, keeping as a constant the heat recovery water flow rate and also its inlet temperature as much as possible. Consequently, Equations (6) and (7) show the electric and thermal efficiency correlations, respectively. The heat-electric power ratio is calculated in Equation (8).

$${\eta }_E=a_1+a_2{P}_E+a_3{P}_E^2+a_{4}Z+a_5{Z}^2+a_6{P}_{E}Z+a_7{P}_E{Z}^2+a_8{P}_E^{2}Z+a_9{P}_E^2{Z}^2$$

(6)

$${\eta }_T=b_1+b_2{P}_E+b_3{P}_E^2+b_{4}Z+b_5{Z}^2+b_6{P}_{E}Z+b_7{P}_E{Z}^2+b_8{P}_E^{2}Z+b_9{P}_E^2{Z}^2$$

(7)

$$R_{HE}=P_T/P_E={\eta }_T/{\eta }_E$$

(8)

Additionally, for diesel, natural gas and air mass flow rate correlations are functions of the electric power and substitution level as shown in Equations (9)–(11), respectively.

$${\dot{m}}_D=c_1+c_2{P}_E+c_3{P}_E^2+c_{4}Z+c_5{Z}^2+c_6{P}_{E}Z+c_7{P}_E{Z}^2+c_8{P}_E^{2}Z+c_9{P}_E^2{Z}^2$$

(9)

$${\dot{m}}_{GN}=d_1+d_2{P}_E+d_3{P}_E^2+d_{4}Z+d_5{Z}^2+d_6{P}_{E}Z+d_7{P}_E{Z}^2+d_8{P}_E^{2}Z+d_9{P}_E^2{Z}^2$$

(10)

$${\dot{m}}_a=e_1+e_2{P}_E+e_3{P}_E^2+e_{4}Z+e_5{Z}^2+e_6{P}_{E}Z+e_7{P}_E{Z}^2+e_8{P}_E^{2}Z+e_9{P}_E^2{Z}^2$$

(11)

2.3. Heat Transfer Considerations

Heat can be transferred from engine coolant and exhaust gases to the heat recovery water. However, in the present study, the heat is transferred only from exhaust gases to heat recovery water, so that the model assumes the heat transfer is proportional to the difference in temperature between the heat recovery water control volume and the combustion engine control volume, which in this case is replaced by exhaust gas temperature. Then, heat transferred is quantified by using an overall heat transfer coefficient as Equation (12) shows [25].

$$Q_{HX}={UA}_{HX}\left(T_{exh}-T_{w,o}\right)$$

(12)

where ${UA}_{HX}$ is the overall heat transfer coefficient between HRWCV and CECV, $T_{exh}$ is the exhaust gases temperature and $T_{w,o}$ is the heat recovery water outlet temperature. It assumes that the lost heat from the engine unit is proportional to the temperature between CECV and the surroundings, thus lost heat is obtained by using another overall heat transfer coefficient, as Equation (13) shows.

$$Q_L={UA}_L\left(T_{ce}-T_{surr}\right)$$

(13)

where $T_{ce}$ is the combustion engine control volume temperature and $T_{surr}$ is the surroundings temperature.

2.4. Combustion Engine Control Volume

The dynamic thermal behavior of a micro-cogeneration system is determined by the engine block thermal mass, encapsulated working fluid and internal heat transfer. A detailed model demands to characterize the individual thermal response of each component above listed, but this model allows us to represent those as a single control volume. Hence, the thermal energy stored within CECV is quantified by using an overall thermal capacitance ${\left[MC\right]}_{ce}$ based on a general energy balance via Equation (14) [33,34].

$${\left[MC\right]}_{ce}{\displaystyle \frac{d{T}_{ce}}{dt}}={UA}_{HX}\left(T_{w,o}-T_{exh}\right)+{UA}_L\left(T_{surr}-T_{ce}\right)+P_T$$

(14)

2.5. Heat Recovery Water Control Volume

Additionally, the model assumes that elements of HRWCV can be represented by using a single overall thermal capacitance ${\left[MC\right]}_w$ that quantifies the thermal energy stored within based on an energy balance via Equation (15).

$${\left[MC\right]}_w{\displaystyle \frac{d{T}_{w,o}}{dt}}={\left[\dot{m}{c}_p\right]}_w\left(T_{w,i}-T_{w,o}\right)+{UA}_{HX}\left(T_{exh}-T_{w,o}\right)$$

(15)

where ${\left[\dot{m}{c}_p\right]}_w$ is the thermal capacitance for heat recovery water mass flow rate and $T_{w,i}$ is the heat recovery water inlet temperature.

3. Experimental Setup

A four cylinder, four stroke, water-cooled, naturally aspired, mechanical direct injection diesel engine was operated as a dual fuel engine with natural gas as primary fuel. The technical specifications of the experimental test engine are given in

Table 1. Technical engine specifications.

Characteristic	Value
Designation	John Deere 4039D DI
Cylinders number	Four (4) cylinders
Cooling type	Water cooled
Charge aspiration	Naturally aspired
Displacement	3.9 L
Compression ratio	17.8:1
Bore × Stroke	106.5 mm × 110 mm
Rated power	41 kW @ 1800 rpm

. The generation unit was linked to an electric resistor load bank with a capacity of 18 kW, an electrical network analyzer type EBCHQ 54115/PD76-24E-WFF was used to measure electric power with a three-phases power controller type NOVUS PCWE-3P-100.

Natural gas as primary fuel had the properties as shown in

Table 2. Natural gas (NG) composition and properties.

Species	${\mathbf{CH}}_4$	${\mathbf{C}}_2{\mathbf{H}}_6$	${\mathbf{C}}_3{\mathbf{H}}_8$	${\mathbf{C}}_4{\mathbf{H}}_{10}$	${\mathbf{CO}}_2$	${\mathbf{N}}_2$
Vol. Comp.	$84.70$	$8.820$	$3.110$	$1.050$	$0.740$	$1.580$
Mass Comp.	$71.54$	$13.97$	$7.224$	$3.215$	$1.719$	$2.335$
Description			Symbol	Units		Value
NG gas constant			$R_{NG}$	$\mathrm{kPa}·{\mathrm{m}}^3/\mathrm{kg}·\mathrm{K}$		$0.4389$
Molar mass			$M_{NG}$	$\mathrm{kg}/\mathrm{kmol}$		$18.943$
Stoich. fuel-air ratio			${\varphi }_{NG,stq}$	$\mathrm{kgNG}/\mathrm{kgAir}$		$0.0619$
Specific heat ratio			${\gamma }_{@{25}^{\circ }C}$	-		$1.2760$
Lower heating value			$LH{V}_{NG}$	$\mathrm{kJ}/\mathrm{kg}$		47,249

. Additionally, commercial diesel was used as a pilot fuel with a lower heating value $\left(LH{V}_D\right)$ of 43,200 kJ/kg and a stoichiometric fuel–air ratio $\left({\varphi }_{D,stq}\right)$ of $0.064935\mathrm{kgDiesel}/\mathrm{kgAir}$. The natural gas mass flow rate was controlled and measured by a sonic flow meter reaching choked flow conditions by flowing through convergent nozzles, ensuring supply conditions such as pressure and temperature. The supply pressure was measured by a gauge pressure electronic transducer type Novus NP-430D with range 0–10 bar and temperature by thermocouple type K. The diesel mass flow rate was measured by using a gravimetric method with a high precision electronic weighing scale type Shimadzu TX3202L. Air mass flow rate was estimated measuring the intake pressure by using an absolute pressure electronic transducer type Wika A-10 with range 0–6 bar, and intake temperature via thermocouple type K.

For the heat recovery water loop, the water mass flow rate was measured by a hall effect water flow meter sensor type YF-S201 with operation range between 1–30 L/min. The heat exchanger implemented to recover heat from exhaust gases was a compact-type build in aluminum, one side with elliptical tubes and the second side with thin fins. Additional temperatures were measured by thermocouples type K, such as heat recovery water inlet and outlet temperature, exhaust gas temperature, engine water coolant temperature and engine oil temperature. Engine block temperature was measured by using an infra-red thermometer type Fluke 568. A detailed schematic diagram of the micro-cogeneration device setup is shown in Figure 2.

The schematic diagram shows a secondary heat exchanger, which dissipates the heat recovered by the primary heat exchanger, allowing it to keep the heat recovery water inlet temperature as constant between 28–33 °C, thus eliminating such temperatures from model equations. At the same time, the heat recovery water flow rate was kept constant and close to 25 L/min ($1.53{\mathrm{m}}^3/\mathrm{h}$ or $425\mathrm{g}/\mathrm{s}$ in mass). It must be clear that heat is recovered only from wasted heat from exhaust gases without heat from the engine cooling system. Finally, the micro-cogeneration system was tested with different electric power generations, such as $2.5$, $7.5$, $10.6$, $14.4$ and $17.7$ kW, at several substitution levels. Electric power generations of 9 kW, $12.5$ kW, $14.4$ kW and 16 kW were then tested with random substitution levels to determine the prediction accuracy of the micro-cogeneration system.

4. Operation Methodology

4.1. Combustion Engine Requirements

According to electric power demand for a continuous operation (primer power), i.e., operation without intermittent under unlimited number of hours, an appropriated electric generator for medium and heavy duty has been selected. These medium and heavy duty electric generators operate normally at 1800 rpm (60 Hz) for a local electric network where dual fuel ICE-based micro-CHP systems would be installed.

Based on the previous, the engine was couple to the electric generator, being designed to operate at the same engine speed so that maximum electric power demand can be supplied. The maximum demanded electric power was set at 30 kW, then $10\%$ was set as over electric power generation, which results in a total of 33 kW. Additionally, knowing that a dual fuel ICE-based micro-CHP system would be installed in a city 1500 m above sea level having an atmospheric pressure of $84.7$ kPa, the altitude power losses must be estimated to achieve the maximum demanded electric power, including electric generator inefficiency.

Rated power and partial load power were corrected according to standard SAE J1349 [35,36]. This method provides equations to calculate the deviations in inlet air and fuel supply conditions. In this study, only inlet air deviations were considered (altitude variation) since there are no density or viscosity values for the local diesel fuel. Corrections were made by Equations (16)–(18).

$$Q=\mathrm{120,000}{\displaystyle \frac{{\dot{m}}_{dm}}{V_{d}N}}$$

(16)

$$f_m=\left\{\begin{array}{cc}0.2,\hfill & \mathrm{if} Q/R<37.2\hfill \\ 0.036(Q/R)-1.14,\hfill & \mathrm{if} 37.2<Q/R<65\hfill \\ 1.2,\hfill & \mathrm{if} Q/R>65\hfill \end{array}\right.$$

(17)

$${P_b}_c={P_b}_r{F}_a={P_b}_r{\left[{\left({\displaystyle \frac{P_{atm}}{99}}\right)}^{\alpha }{\left({\displaystyle \frac{298}{T_a+273}}\right)}^{\beta }\right]}^{f_m}$$

(18)

Based on previous equations and looking for the commercial stationary engine brands at the installation city, Cummins and John Deere were the best options according to factors such as quality, reliability, price and availability of spare parts. Several available options were reviewed, finding the best option to John Deere 4039D, to which the SAE J1349 was applied, obtaining the results as

Table 3. Values for electric power generation corrections.

Description	Symbol	Units	Value
Cylinder capacity	$V_d$	L	3.9
Engine speed for maximum power	N	rpm	1800
Declared fuel consumption	${\dot{m}}_{dm}$	L/h (g/s)	10.5 (2.48)
Pressure ratio *	R	−	1
Pressure exponent *	$\alpha$	−	1
Temperature exponent *	$\beta$	−	0.7
Altitude correction factor	$F_a$	−	0.944
Electric generator efficiency	${\eta }_{eg}$	−	0.88–0.92
Prime rated power (PRP)	${P_b}_r$	kW	41.0
Corrected power	${P_b}_c$	kW	38.7
Maximum electric corrected power	${P_E}_c$	kW	34.8

* Applied values for naturally aspired engines.

shows.

Due to altitude and electric generator losses, maximum electric power is $34.8$ kW as shown in Table 3, which is barely higher than the maximum demanded electric power of 33 kW (even with $10\%$ of over electric power). This engine has a built-in automatically controlled diesel fuel system by centrifugal weights that allows us to keep the engine speed at 1800 rpm.

4.2. Conventional Operation as Reference for Dual Operation

At first, it was needed to measure a conventional diesel operation map, which means $Z=0$, measuring variables such as fuel consumption, exhaust gases flow and temperatures. By knowing the conventional fuel consumption at several electric powers, then a natural gas flow rate can be initially estimated to operate the micro-cogeneration system at a desired substitution level and fixed electric power. Fuel consumption, when operated with conventional diesel, is shown in Figure 3. Inputs such as electric power and substitution level are entered into a computer interface that controls electronic actuators and read sensors for several variables.

Thereby, a equation for the tendency fitted curve can be obtained for diesel fuel mass flow rate at conventional diesel operation as a function of electric power generation, as Equation (19) shows. When natural gas is introduced to the intake system, a portion of the mass flow rate of diesel and its energy has to be substituted, but instead natural gas has to add approximately the same amount of energy what was substituted from diesel, if it is assumed that electric efficiency is kept as constant. The previous statement allows us to obtain Equation (20), which combines Equation (1) as well.

$${\dot{m}}_{D,c}=1.657P_E^2+13.210P_E+759.68$$

(19)

$${\dot{m}}_{NG}={\dot{m}}_{D,c}Z{\displaystyle \frac{LH{V}_D}{LH{V}_{NG}}}$$

(20)

Equation (20) tends to estimate a natural gas mass flow as a function of electric power generation and substitution level. As said above, it considers no electric power inefficiency changes due to variations of substitution level increasing, that is why, during micro-cogeneration system operation, both the desired substitution level and the actual substitution level can be seen, where the last one is calculated from the actual natural gas mass flow measured in real time, allowing it to be closer to desired value.

4.3. Heat Exchanger Selection Criteria

Considerations for heat exchanger selection have been taken such as a lower pumping work in the exhaust gases side to reduce back pressure on the engine, a large heat transfer area but with a short heat exchanger volume, a low cogeneration system installation area, existing design on the local market and low cost related to manufacturing and maintenance.

Based on previous considerations, a compact heat exchanger type cross flow tube-fin with both fluids unmixed seemed the best selection due to its large free-flow area for the exhaust gases passing through fins, high ratio between total heat transfer area and heat exchanger volume, generally given for these arrangements of heat exchangers with ratios higher than $700{\mathrm{m}}^2/{\mathrm{m}}^3$), being highly commercial in the local market for automotive industry which means a low cost related to manufacturing, installation and maintenance as well. Therefore, selection criteria values for heat exchanger were taken from initial experimental data for conventional diesel operation, trying to obtain the maximum exhaust gases flow rate and its respective temperature, also by setting a desired heat recovery water flow rate and corresponding inlet and outlet temperatures as is shown in

Table 4. Initial experimental data for conventional diesel and desired temperatures.

Description	Symbol	Units	Value
Inlet heat recovery water temperature	$T_{w,i}$	° C	20
Outlet heat recovery water temperature	$T_{w,o}$	° C	70
Exhaust gases temperature (average)	$T_{exh}$	° C	285
Exhaust gases mass flow rate	${\dot{m}}_{exh}$	kg/s	0.239
Heat recovery water mass flow rate	${\dot{m}}_w$	kg/s	0.225
Heat recovery water volumetric flow rate	${\dot{V}}_w$	L/min	13.62

.

Then, using values in Table 4, initial fluid properties can be estimated for both heat recovery water and exhaust gases, which then allows us to apply an iterative method to obtain dimensionless characteristic numbers and convective heat transfer coefficients for both liquid and gas sides [37]. Subsequently, by having previous heat transfer coefficients and heat transfer conductance coefficients for material, a total heat transfer coefficient can be obtained for calculating a Number of Transfer Units symbolized $NTU$. The method proposed by Kays and Crawford, for an unmixed cross flow heat exchanger [38], is shown in Equation (21), also methods to determine values such as effectiveness $\epsilon$ and pressure losses $\Delta p$ can be found.

$$\epsilon =1-exp\left({\displaystyle \frac{NT{U}^{0.22}}{c}}\left[exp\left(-c·NT{U}^{0.78}\right)-1\right]\right)$$

(21)

Finally, based on the above-described methodology, the general characteristics of the selected compact heat exchanger are shown in

Table 5. Heat exchanger characteristics and estimations.

Description	Symbol	Units	Value
Heat exchanger volume	$V_t$	m3	0.0018
Heat transfer area-volume ratio	${\alpha }_f$	m2/m3	1474
Total tubes	$n_t$	-	29
Total fins	$n_f$	-	11,100
Tubes side pressure drop	$\Delta p_t$	kPa	17.38
Fins side pressure drop	$\Delta p_f$	kPa	11.37
Effectiveness	$\epsilon$	-	0.331
Heat recovery potential (ideal)	$Q_{max}$	W	67,242
Actual possible heat recovery	$Q_a$	W	22,195

. Additionally, estimated values for heat recovery system operation are shown in Table 5, such as thermal recovered power, heat exchanger effectiveness and even pressure losses for both sides.

As shown in Table 5, a high ratio between total heat transfer area and heat exchanger volume was achieved close to 1474 m²/m³, more than twice the common values. Additionally, an estimation of actual possible heat recovery of 22.2 kW, offers a high potential for combined power generation, both electric power as much as heat power, having low pressure drop for tubes sides and low back pressure is generated by heat exchanger to the engine (17.38 kPa), allowing us to conserve as high as possible electric power generation efficiency.

4.4. Gaseous and Pilot Fuel Supply Control

The natural gas fuel flow is controlled by a sonic flow meter, which operates as a function of supply pressure and temperature, fixing values for nozzle diameters and discharge coefficients. Knowing that supply fuel temperature was kept constant, the natural gas mass flow rate depends only on supply pressure [39]. The previous procedure can be inverted, so that supply pressure can be obtained by a desired natural gas flow rate, then by combining Equation (20) on actual choked flow rate equations, Equation (22) is obtained as a result.

$$p_0={\dot{m}}_{D,c}Z{\displaystyle \frac{1}{A^{*}{C}_d}}{\displaystyle \frac{LH{V}_D}{LH{V}_{NG}}}\sqrt{{\displaystyle \frac{R_{NG}{T}_0}{\gamma }}}{\left({\displaystyle \frac{\gamma +1}{2}}\right)}^{{\displaystyle \frac{\gamma +1}{2\left(\gamma -1\right)}}}$$

(22)

where $A^{*}$ is the nozzle critical area and $C_d$ its discharge coefficient. Thereby, the natural gas flow rate can be controlled depending only on electric power and substitution level. For this study, supply absolute pressure was fixed at a maximum of $7.3$ bar, for reducing possibilities of NG leakages and safety work. Minimum supply absolute pressure was fixed at $1.6$ bar, to warranty chocked (sonic) flow conditions. In parallel, four nozzles were installed and computationally controlled.

Table 6. Sonic nozzle flow meter operation parameters.

Throat Diameter	${\mathit{C}}_{\mathit{d}}$	Operation Range
$0.4064$ mm ($0.016$ in)	$0.8853$	$0<Z<15$
$0.6096$ mm ($0.024$ in)	$0.8982$	$15\le Z<25$
$0.8890$ mm ($0.035$ in)	$0.9219$	$25\le Z<55$
$1.3208$ mm ($0.052$ in)	$0.8999$	$55\le Z<70$

shows the substitution level operation range for each nozzle independently of electric power generation, then Figure 4 shows the operation map of the sonic nozzle flow meter for $0.8890$ mm ($0.035$ in) throat diameter for a substitution level range from $25\%$ up to $55\%$.

In both conventional and dual operation modes, the diesel fuel flow is controlled automatically by an automatic mechanical system of weights built inside the engine that allows us to keep the engine speed at 1800 rpm.

4.5. Control and Visualization Interface

Devices and equipment explained previously were linked by using a National Instruments data acquisition and control system. A real-time operation was allowed to modify operation parameters such as orifice selection for natural gas fuel flow, safety assurance of sonic flow meter, expected substitution level Z, visualization of working parameters of engine and heat recovery systems such as temperatures, diesel fuel flow, water flow, power and efficiency at operation, and also data recording. A screenshot of the Labview interface is shown at Figure 5.

5. Results and Discussion

The micro-cogeneration system operation maps were obtained in dual fuel mode at several substitution levels, keeping the water heat recovery mass flow rate and inlet temperature as constant, at steady state, thus, the models for diesel, natural gas and air mass flow rate, electric and thermal efficiency are valid at the same state, i.e., steady operation for a long time. Nonetheless, other results are presented when running at transient operation for a desired change of thermal or electric output power or even heat recovery water outlet temperature. Previous studies have calibrated the models for micro-cogeneration systems with second- or third-degree polynomials as a function of electric power depending on the desired fit with experimental data [34,40]. However, those models presented in previous studies do not include the substitution level as a factor, since they only operate with one fuel. The coefficients obtained for calibrated models regarding the electric and thermal efficiencies as well as diesel, natural gas and air mass flow rate are shown in

Table 7. Coefficients for model correlations.

Coefficients	${\mathit{\eta }}_{\mathit{E}}$	${\mathit{\eta }}_{\mathit{T}}$	${\dot{\mathit{m}}}_{\mathit{D}}$	${\dot{\mathit{m}}}_{\mathbf{NG}}$	${\dot{\mathit{m}}}_{\mathit{a}}$
−	$0.006248$	$0.331000$	$719.853$	$-0.79160$	50,139.1
$P_E$	$0.029212$	$-0.004081$	$21.2416$	$0.0$	$-110.647$
$P_E^2$	$-0.000832$	$-0.000001$	$1.44269$	$0.0$	$0.0$
Z	$0.101548$	$-1.011470$	$-549.056$	$0.0$	−14,331.6
$Z^2$	$-0.741322$	$1.766230$	$0.0$	13,943.5	17,331.9
$P_{E}Z$	$-0.071513$	$0.163051$	$-55.3363$	$387.704$	$475.929$
$P_E{Z}^2$	$0.148371$	$-0.299428$	$0.0$	$-2031.57$	$-537.707$
$P_E^{2}Z$	$0.004091$	$-0.006119$	$0.0$	$-19.8360$	$0.0$
$P_E^2{Z}^2$	$-0.007169$	$0.011386$	$0.0$	$78.1700$	$0.0$

.

As shown in Table 7, all correlations are functions of the electric power and substitution level. It is noted that no coefficients for a overall efficiency correlation are presented, but it must be quantified as the arithmetic sum of the two efficiencies. A similar procedure occurs for the electric and thermal power ratio as shown in Equation (8).

5.1. Steady State Model Results

The operation maps were obtained by testing the micro-cogeneration system to several electric powers, however, operation data for electric power of 9 kW, $12.5$ kW, $14.4$ kW and 16 kW were excluded out of model calibration, aiming to prove the prediction accuracy with them. Figure 6, Figure 7 and Figure 8 show the results obtained by models (continuous lines) and experimental data (markers) for diesel, natural gas and air mass flow rates in steady state for micro-cogeneration system operation, respectively.

Figure 6 shows that the air mass flow rate decreases when the substitution level increases, which is because the substitution level increases the natural gas flow rate displacing a volume portion of the engine intake air for combustion. At a constant substitution level, the air mass flow rate continues to decrease due to a higher electric power produces an intake temperature increasing [31,41]. Other important behavior occurs when the substitution level moves between $30\%$ to $50\%$, holding the air mass flow rate as constant and close to 47.3 g/s. The result for natural gas mass flow rate is shown in Figure 7.

Figure 7 shows that the natural gas flow rate increases when the substitution level grows, because natural gas substitutes an important portion of the chemical energy supplied by diesel. However, an opposite effect occurs when the substitution level remains constant but electric power varies, having higher natural gas flow rates for lower electric powers, which can be associated with electric efficiency reductions when the engine operates at low loads. On the other hand, the diesel mass flow rate presents an expected tendency according to original engine operation as conventional diesel, having lower diesel mass flow rates when both electric power and substitution level decrease, as shown in Figure 8. From the above, engine electric power inefficiencies were compensated by the natural gas mass flow rate increasing. Knowing the incoming flows to engine intake system, the equivalence ratio can be obtained as shown in Figure 9. The equivalence ratio results agree with reported values on several studies of dual fuel diesel engines, increasing the values when higher electric power is demanded [17].

Now, the results for electric and thermal efficiencies are shown in Figure 10 and Figure 11, respectively. For higher electric powers, the higher electric efficiency is achieved. The highest value for electric efficiency was close to $27\%$, but had a pronounced decreasing until $20\%$ when the substitution level increases from $20\%$ up to $60\%$. For electric power from $2.5$ kW up to $14.4$ kW, maximum values of electric efficiency were achieved for conventional diesel engine operation (i.e., $0\%$ of substitution level) [42,43], then decreasing linearly when substitution level grows. Figure 11 shows the results for thermal efficiency, substitution level has a slight effect on thermal efficiency decreasing, having values from $24.5\%$ up to $30\%$ at electric power from $7.5$ kW up to $17.7$ kW. A thermal efficiency atypical behavior is obtained at $2.5$ kW of electric power, with the highest value close to $32\%$, but also with fast decreasing close to $23\%$ when the substitution level grows up to $25\%$.

As a combination of electric and thermal efficiencies, the overall efficiency is shown in Figure 12, which shows a matching behavior to the electric efficiency shown in Figure 10, so that a higher overall efficiency is obtained for higher electric power, presenting marked drops for electric powers such as $2.5$ kW, $7.5$ kW and $10.6$ kW, but less marked decreases for $14.4$ kW and $17.7$ kW of electric power. The maximum overall efficiency was close to $52\%$. Then, the results for heat-electricity ratio are presented in Figure 13, showing that higher heat is recovered, relative to electric power, for lower electric power, presenting values between from 4 to 6 in case of $2.5$ kW, meanwhile at higher electric power, the heat-electricity ratio tends to keep from 1 up to 2, i.e., the thermal power recovered is equal to electric power generation.

5.2. Steady State Model Random Validation

Trying to know the model accuracy at a random micro-cogeneration system operation, depending on electric power generation and natural gas availability that require changes of the substitution level, the micro-cogeneration system was operated at several random operation points. In that way,

Table 8. Random operation points for model validation.

Operation Point	1	2	3	4	5	6	7	8
$P_E$ [kW]	9	$12.5$	$12.5$	$14.4$	$14.4$	$14.4$	16	16
Z [-]	$0.00$	$0.00$	$37.2$	$36.9$	$24.8$	$53.8$	$0.00$	$34.3$

lists the random operation points and their respective electric power and substitution level achieved. It must be known that random operation points were not used to predict model coefficients nor model calibration.

Thus, a comparison between experimental results and model predicted values can be shown with respective errors. Figure 14 shows efficiency prediction and measured values, and the maximum error was close to $20\%$ at $12.5$ kW of electric power generation and $37\%$ of substitution level for thermal efficiency, which agrees with the results in Figure 11, where thermal efficiency presents the largest differences between predicted and measured value. Overall efficiency presents a similar trend because it is affected by thermal efficiency behavior but having a maximum error that is slightly lower, around $15\%$. In the case of electric efficiency, the error remains lower than other efficiencies, with an average values close to $5\%$.

Additionally, for mass flow rates such as diesel fuel and natural gas, and intake air, a comparison between experimental results and model predicted values are shown with their errors on Figure 15 and Figure 16, respectively. For the mass flow rate of fuels, maximum errors are around $9\%$ and $7\%$ for diesel and natural gas, respectively. However, along all random operations, the average error was close to $2\%$ for both fuels. Regarding to air mass flow rate, accurate results were obtained, having a maximum error around $1.5\%$ for the first operation point, but then going lower than $0.5\%$.

5.3. Transient State Model Validation

Micro-CHP systems can offer wide applications, as required water outlet temperature $T_{w,o}$ in industry, heating processes, also water flow rate ${\dot{m}}_w$ could change as well, so that transient validation aims to know the predictability for water outlet temperature. Additional temperatures were measured in micro-cogeneration system operation as well as steady state operation, as Figure 17 shows. Additionally, such operation points are shown in

Table 9. Operation points for measured temperatures.

Operation Point	1	2	3	4	5	6	7	8
$P_E$ [kW]	$2.5$	$7.5$	$10.6$	$14.4$	$17.7$	$2.5$	$2.5$	$7.5$
Z [-]	0	0	0	0	0	$16.6$	$23.2$	$24.2$
Operation Point	$\mathbf{9}$	10	$\mathbf{11}$	$\mathbf{12}$	$\mathbf{13}$	$\mathbf{14}$	$\mathbf{15}$	$\mathbf{16}$
$P_E$ [kW]	$7.5$	$10.6$	$14.4$	$14.4$	$14.4$	$17.7$	$17.7$	$17.7$
Z [-]	$28.6$	$30.6$	$24.8$	$36.9$	$53.8$	$24.6$	$45.9$	$60.7$

.

Energy balances are obtained for each control volume as it is shown by Equations (12)–(15) for any operation state, i.e., steady and transient of the micro-cogeneration system. It is necessary to keep in mind that, in the present study, only heat from exhaust gases was recovered, excluding the heat from engine water coolant, which allows us to obtain the above equations. Applying measured temperature values from Figure 17 on the above equations assuming steady state can be obtained results for both overall heat transfer coefficients ${UA}_{HX}$ and ${UA}_L$, and even correlation as a function of electric power generation and substitution level. Thus, Equation (23) can be obtained, representing the calibrated model for overall heat transfer coefficient between HRWCV and CECV.

$$\begin{array}{cc}\hfill {UA}_{HX}=& 0.078468-0.003424P_E+0.000133P_E^2-0.137199Z+0.377765Z^2\hfill \\ \hfill & +0.035587P_{E}Z-0.080694P_E{Z}^2-0.001662P_E^{2}Z+0.003437P_E^2{Z}^2\hfill \end{array}$$

(23)

Transient state validation for micro-cogeneration system operation was carried out by comparison between experiments and modeling results, applying a step input of two random electric power levels for the combustion engine, corresponding from 9 kW to $12.5$ kW. Data for each variable were collected per second during states as steady–transient–steady. The overall thermal capacitance ${\left[MC\right]}_w$ for HRWCV was equal to $180\mathrm{kJ}/\mathrm{K}$. Therefore, heat recovery water outlet temperature $T_{w,o}$ behavior can be obtained for both measured and modeling results, which are shown in Figure 18. Additionally, measured inlet water heat recovery temperature $T_{w,i}$ and exhaust gas temperature are shown in Figure 18.

Figure 18 shows the modeling result for the heat recovery water outlet temperature, which fits to measured values even when it is used as a constant overall heat transfer coefficient ${UA}_{HX}$ for a determined electric power and a substitution level by using Equation (23). Differences are obtained just at the beginning of the input steps, but such a difference decreases when the micro-cogeneration system is achieving the steady state operation.

6. Conclusions

For experimentation, it is necessary to properly choose the heat exchanger system to keep the lowest engine back pressure to hold the engine thermal efficiency as high as possible, low pumping energy, low pricing and maintenance ratio, high heat recovery properties, and no contamination of heat recovery water for possible uses in processes. Thus, compact heat exchanges contain enough features over other types of heat exchangers as shell and tubes, and even some ejectors considered in the first stage of study. The heat exchanger selected in this study offered pressure drops of $17.38$ kPa ($0.174$ bar) and $11.37$ kPa ($0.114$ bar) for the tube and fin sides, respectively. There was a high value for heat transfer area and volume ratio of $1474{\mathrm{m}}^2/{\mathrm{m}}^3$ with no crossing flow to keep clean water.

In this study, thermal isolation of a engine exhaust system and exhaust gases diffuser was applied by using ceramic fiber blanket. The length of the engine exhaust gases system is close to $2.5$ m, which induced high thermal energy losses prior to the heat exchanger.

A gravimetric method by using a weighing scale for liquid fuel allows an easier and precise measurement task, even though the engine has a large diesel flow return from the injection system to fuel tank, due mechanical pumps. However, precise ground level and vibration isolation must be assured for the weighing scale. On the other hand, for the measurement and supply of NG fuel flow, the sonic flow meter allows us to reduce the operation time at dual mode even without automatic close loop control. High accuracy and large mass flow range is featured adjusting only the gas supply pressure as probed in previous studies. The sonic flow meter can be improved by implementing electronic pressure control valves coupled to a PID strategy.

Now, regarding the thermal performance of dual fuel diesel engines, it lets to reduce the electric power generation costs related with heavy fossil fuel consumption as diesel being replaced by natural gas using or even biogas from local sources, more diesel replacement represents a higher substitution level for engine operating. The highest substitution level was around $60\%$ at an electric power generation of $17.7$ kW. However, higher substitution levels (higher than $60\%$) were limited by the electric power generation control system, which does not allow us to achieve a combustion engine rated power or electric power generation greater than $17.7$ kW. The maximum substitution level was $23\%$ at the lowest electric power generation of $2.5$ kW, but presenting an unstable operation of the cogeneration system followed by the lowest electric efficiencies of $4\%$.

A higher exhaust gas temperature is achieved for higher substitution levels; however, thermal efficiency decreases by small amounts. Thermal efficiency decreases by larger amounts when increasing the electric power generation compared with substitution level increases. The highest thermal efficiency was $33\%$ at the electric power generation of $10.6$ kW and substitution level of $40\%$. The highest electric efficiency was $28\%$ at the electric power generation of $17.7$ kW and substitution level of $24\%$. The overall efficiency presents similar behavior, reaching its maximum value of $53\%$ at $14.4$ kW and a substitution level of $0\%$. Recovering heat from engine water coolant could allow us to reach higher overall efficiency for cogeneration system, some studies report values close to $75\%$.

Additionally, the maximum error presented was close to $20\%$ associated to thermal efficiency. However, errors for all other variables were lower than $10\%$ for most of the cogeneration system operation points. The modeling transient result for outlet water heat recovery temperature fits to the experimental measuring results. Therefore, calibrated empirical correlations make it possible to control a cogeneration system depending on both demand for electric power generation or thermal power requirements accurately.

Finally, for the next stages of this study, the authors want to implement not only heat from exhaust gases, but also from an engine cooling system, with an electronic close loop control for the water temperature and substitution level as set-points based on NG availability, likely using electronically controlled injection system as a common rail for the engine, allowing us to change the injection timing aiming to improve both electric and thermal efficiencies, and also allowing us to achieve higher substitution levels even at low electric powers.