E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.

1051/e3sconf/20198501012
EENVIRO 2018

Energy analysis of a CHP plant with internal combustion

engines, for a district heating system, based on the
information from the annual database
Gabriel Mărcuș1, Vlad Iordache1, Florin Iordache1, and Anica Ilie1
1CAMBI Research Center, Technical University of Civil Engineering of Bucharest, Romania

Abstract. Building a model for consumption, production and efficiency of a combined heat and power
(CHP) system can bring important data for planning and management activity of such a technological
system. The results of this study will show the overall efficiency of a real cogeneration plant over a
year. The study is based on the information from the daily database of an economic operator, during
2012. The CHP plant, having reciprocating internal combustion engines (RICE) as prime movers,
provides the thermal energy to the district heating system of a city from Romania with 129,368
habitants. RICE are operating in simultaneous mode or in partial load. The numerical model reveals
the behavior of the daily thermal, electrical and global efficiency, accordingly to the partial load. The
model was applied in both calculation assumptions: using the lower heating value and the higher
heating value of the natural gas. A statistical analysis of efficiencies of the CHP plant was made. Was
performed the statistical analysis of the database efficiency (also called real efficiency) with the global
efficiency calculated by the model (also called modelled efficiency). Linear and multiple regression
equations explain the variance of the real efficiency and of the modelled global efficiency.

useful electrical power compared to the other modes of

1 Introduction operation.
In order to continue the scientific research of the
Cogeneration systems involve the combined production
processes occurring in a CHP plant with RICE,
of electricity and heat for space heating and hot water,
information were obtained and analysed from an
using a single source of fuel. Cogeneration technologies
economic operator database. This economic operator
comprise several subsystems and equipment with multiple
supplies the thermal energy to a city in Romania with
interconnection capabilities. In the case of classical
129,368 inhabitants (year 2012). The numerical model
systems, the efficiency of separate heat and electricity
built for the cogeneration plant, based on the daily data
production is, accordingly to the literature, about 58%.
information had the ultimate goal the comparison between
In the case of cogeneration the efficiency reaches 85% [1].
the daily global efficiency calculated with previously
Cogeneration technology based on (RICE) enjoys wider
mentioned formulas, named 'modelled efficiency' in the
appreciation and spreads among investors' decisions at the
present paper, and the daily efficiency from the database,
time of implementing the method of producing the two
named 'real efficiency' in the present paper. The daily
forms of energy, heat and power.
thermal, electrical and global values of efficiency for the
In a previous study [2], the CHP plant efficiency was
whole of 2012 are processed and analysed statistically in
compared for three different operation modes.
relation to the partial load, PL, accordingly to the daily
This study used:
operating hours recorded in the database. Variations in
• the equation found by Sanaye S. et al. [3] regarding efficiency trends were observed and linear regression
the thermal efficiency at partial load, equations have been developed for thermal, electrical and
• ASHRAE equation 2008 [4] regarding the variation global efficiency. The slope of regression line was found
of the efficiency of the electric generators and and the intercept for PL=50% (the minimum partial load).
• the method of calculating the energy requirements The present study compares the modelled efficiency
and the system efficiency in SR EN 15316 [5], [6]. values with the real efficiency values from two
The conclusion was that the simultaneous operating perspectives: the calculation using lower heating value
mode of engines at the same load without the boiler and the calculation using higher heating value. At the end
intervention is the most advantageous. Simultaneous of this work is elaborated a multiple regression equation
mode of operation has led to: obtaining maximum for the real efficiency which takes into account the partial
individual thermal efficiency on wider ranges of load load and the modelled efficiency.
values, observing shorter maintenance periods due to the The method described in this paper makes evaluation
shorter periods of use of the boiler and getting 40% more possible to estimate precisely what global efficiency is
*
Corresponding author: [email protected]

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (http://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

really obtained in a cogeneration plant similar to the plant Note that for other types of engines, the model cannot be
under consideration. The method can be extrapolated, applied using the same database information used in the
under certain conditions, to systems made up of several current paper simply by changing the technical
RICE, with the same technical characteristics (the same specification entry; it requires new recorded data and a
manufacturer). The numerical model can be applied to new experiment with a time step that may be different
other types of RICE-s by introducing in the model both: from the daily one. The conclusions will be more accurate
the corresponding technical specifications and the as the registration time step will be reduced.
production and consumption parameters recorded during
periods of operation, for these new types of equipment.

Nomenclature M1 first reciprocating internal combustion engine

Ḃ daily mass flow of natural gas (m3N/day]) M2 second reciprocating internal combustion engine
PLCOG partial load of cogeneration system (%) G1 first electrical generator
PL partial load (%) G2 second electrical generator
LHV lower heating value (9.166 kWh/m3N)
HHV higher heating value (10.834 kWh/m3N) Greek letters
Eel,G1 daily electricity suppled by the first generator ηelCOG modelled electrical efficiency (%)
(MWh/day) η2E
elCOG modelled electrical efficiency without electrical
Eel,G2 daily electricity suppled by the second generator outliers values (%)
(MWh/day) ηthCOG modelled thermal efficiency (%)
Qth,M1+M2 daily heating energy suppled by the engines η2E
thCOG modelled thermal efficiency without thermal
(MWh/day) outliers values (%)
CHP Combined heat and power ηglobal modelled global efficiency (%)
η2E modelled global efficiency without global
Abbreviations global

Boiler thermal boiler outliers values (%)

RICE reciprocating internal combustion engine ηBDB real efficiency recorded in the database (%)
PHE plate heat exchanger

RICE-s and hot water boilers are connected to plate

2 Technological scheme heat exchangers for heat transmission to the district
The cogeneration system analysed, Fig. 1, has two RICE- heating grid. The water supply for the primary circuits or
s Jenbacher JMS 620 GS-NL equipped with electric for the addition water in case of losses pre-treated in a
generators, with a nominal power of 3119 kW, according water softening/degassing module by reverse osmosis.
to ISO standard (ICFN), used as prime movers for the The fuel for the boiler burners and for the combustion
production of heat and power. The peak source used chambers of the engines is the natural gas from the
consists of two hot water boilers, 35 MW International Prahova county basin with the lower heating value LHV
Loos. For the domestic hot water recirculation system, in = 9,17 kWh/Nm3.
the summer when there is no demand for heating load, a
hot water boiler of 11.6 MW is used.

Fig. 1. Technological scheme of CHP.

2
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

3 Database
The daily recorded data for year 2012 consist in: the September - between 70% ÷ 73%). During the year 2012
operating hours of each of two engines, the daily and there have been several hours when one of the
hourly consumption of natural gas, the daily and hourly cogeneration modules (thermal engine + electrical
electricity supply produced by each of electrical generator + turbo charger) was not operational due to the
generator, the daily and hourly heat supply - cumulative maintenance interventions.
value on the two internal combustion engines - and the The natural gas consumption of the cogeneration
daily global efficiency of the system recorded by the system ranges between a minimum of 17898 m3/day on
operator. Regarding the daily global efficiency recorded January 4, 2012, corresponding to a period when only one
data along the months of 2012, it can be observed that they engine was operational (for the first 7 days of January
have higher values during the winter months (October, 2012) and a maximum of 39373 m3/day on October 27,
November, December, January - around 75% ÷ 79%) and 2012 corresponding to a period when both engines were
smaller values during the summer months (July, August, operational (Fig. 2).

Fig. 2. Consumption of natural gas, October 2012, m3/day.

value, LHV = 9,1667 kWh/Nm3, in which case the

differences from the real efficiency were higher between
4 Numerical model 4.34% and 5.40% and second - considering higher heating
The initial numerical model [2] configured in operating value, HHV = 10,8347 kWh/Nm3, in which case the value
mode 3, which means simultaneous operation at the same differences from the real efficiency were lower by 5.96 %
loads without the intervention of the boiler, is applied to to 9.58 %, Fig. 4. The first case considered was closer to
the cogeneration system engines. The model, Fig. 3, uses reality and was adopted to lead this analysis. Finally, the
the daily time step during the year 2012. From the total thermal efficiency and was closer to reality and was
database were taken the measured data for: the cumulated adopted to lead this analysis. Finally, the total thermal
heat output of the two engines (M1 and M2) and the daily efficiency and the global (thermal + electric) efficiency of
operating hours of each engine, the individual production CHP system have been calculated.
of the electric energy of each engine in and the daily The initial numerical model [2] was built for MWM
global efficiency, named as real efficiency. internal combustion engines with a nominal thermal
In simultaneous operation mode, M1 and M2 divide output of 1675 kW. Accordingly to the analysis
heat production equally during daytime running hours. At performed on the MWM engines, the partial load interval
the division of heat production we had introduced the (PL) when the maximum thermal efficiency is reached
condition of the nominal thermal output of the engine in ranges from 85.77% to 91.47%. It has been a matter of
the data sheet - 3070 kW. In the months of October, checking the optimum PL range for the M1 and M2
November and December the thermal boilers operates to engines of actual plant, with the thermal output of 3070
provide the heating peak load. For each engine was kW. First, the technical data of the new engines were
calculated in the model, after dividing the cumulative heat introduced in the model (the nominal thermal output and
output: the partial load (PLCOG), the individual thermal the nominal electrical power). The maximum thermal
and electrical supply at partial load and the useful energies efficiency was 42.96% on the new partial load intervals,
of entire system: thermal, electrical, total, the lost energy respectively 85.31% - 91.95% PL for M1, and 85.31% -
and the consumed fuel energy by the cogeneration system. 91.97% PL for M2. It can be seen that the higher the
For the fuel energy input, the model was tested in both engines nominal power the wider the maximum thermal
possible situations: first - considering the lower heating efficiency range.

3
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

Fig. 3. The model of energy analysis of database.

Fig. 4. Modelled efficiency and real efficiency with LHV and HHV.

A calculation of the operating time at the maximum The frequency of occurrence of the global daily
thermal efficiency was performed accordingly to the modelled efficiency (in day’s number) is shown in Fig. 5.
optimal operating interval of partial load (85.31% - The values of global daily modelled efficiency ranged
91.97% PL). The highest number of days (17 days) at the between 60 % and 100 % have been investigated, with a
maximum thermal efficiency was in February 2012. 5 percent step. The most frequent occurrence of 85 %
efficiency is observed in 185 days of the year. The

4
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

analysis of daily global efficiency figures has led to the

observation of a wrong recording for two consequently
days (more than 100% global efficiency).

Fig. 5. Frequency of daily global modelled efficiency, 2012.

work the whole day or during the maintenance periods.

Also will be removed the one day wrong recording data
5 Energy analysis (more than 100% thermal efficiency).
The energy analysis has proposed the observation of the The variation of the modelled electrical efficiency,
cloud of points, the identification of the outliers values ηelCOG, depending on the PL of engines assembly, PLCOG,
and the highlighting of the trends of variation of the Fig. 6, b., is almost constant and it can be said that partial
modelled efficiencies. load operation does not influence its variation. At 50% PL
Based on the primary visual analysis, it is observed (the minimum operating load), the electrical modelled
that the modelled thermal efficiency, ηthCOG, exhibits an efficiency in the model is 41.72%, a very small decrease
increasing variation with the partial load of the engines of 1.88% from the maximum electrical modelled
assembly, PLCOG, Fig. 6, a. efficiency (43.60%).
At the same partial operation load, several values of In this case, also the observation is valid: at the same
the modelled thermal efficiency are recorded on different operating PL several values of the electrical efficiency,
days of the year with very small differences between them with small differences between them (vertically close
(vertical points). At the same partial load, the calculated points), are recorded. From the diagram of the normal
daily thermal efficiency may vary to a maximum of 4 distribution of the electrical modelled efficiency, the
percent. points that appeared dispersed (the red points) called
From the diagram of the normal distribution of the 'electrical outliers values' will be removed. The outliers
thermal modelled efficiency, the points that appeared were mainly values recorded during the days when one of
dispersed (the red points) called 'thermal outliers values' engines did not work the whole day or during the
will be removed. The outliers values were mainly maintenance periods. Also will be removed the one day
recorded during the days when one of engines did not wrong recording data (more than 100% electrical
efficiency).

5
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

a. thermal outliers - abnormal operation mode, optimal values - normal operation mode.

b. electrical outliers - abnormal operation mode, optimal values - normal operation mode.

c. global outliers - abnormal operation mode, optimal values - normal operation mode.

Fig. 6. Variation of modelled efficiency as a function of the partial load: a. Thermal efficiency, b. Electrical efficiency, c. Global
efficiency.

Based on the primary visual analysis, the variation of dispersed (the red points) called 'global outliers values'
the modelled global (thermal + electric) efficiency of the will be removed. The outliers were mainly values
cogeneration plant, exhibits an increasing variation with recorded during the days when one of engines did not
the partial operation load of the engines assembly, PLCOG, work the whole day or during the maintenance periods.
Fig. 6, c. Also will be removed the one day wrong recording data
Similar to the pattern of the modelled thermal (more than 100% global efficiency).
efficiency, a concentration of cloud of points with a The graph of the remaining thermal efficiency values
obvious trendline between 72 % and 100 % PL was was plotted, Fig. 7. The regression line makes an 23.42
observed. At the same partial operating load, obviously, degrees angle with the horizontal, having a slope of
several global efficiency values are recorded on different 43.32%. Equation (1) expresses the variation of the
days with very small differences between them (vertically modelled thermal efficiency after the elimination of the
close points). At the same PL, the daily global modelled thermal outliers values.
efficiency returns values that can vary to a maximum of 5 2𝐸𝐸
percent. 𝜂𝜂𝑡𝑡ℎ𝐶𝐶𝐶𝐶𝐶𝐶 = 0.4332 ∙ 𝑃𝑃𝑃𝑃𝐶𝐶𝐶𝐶𝐶𝐶 + 0.0015 (1)
From the diagram of the normal distribution of the
global modelled efficiency, the points that appeared

6
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

Fig. 7. Variation of modelled thermal efficiency without 'thermal outliers values' with partial load - normal operation mode.
The coefficient of determination R2 = 0,9443 shows efficiency, the goal was to find a multiple regression
that in this case the values of the independent variable, equation (4) that mathematically describes how the real
PLCOG, determine in a large percentage of 94,43 % the efficiency (ηBDB ) depends on the partial load (PL) and on
values of the modelled thermal efficiency - the dependent the modelled global efficiency (ηglobal ). The 3d chart was
variable. created, Fig. 10.
Similarly, the variations of the modelled electrical
efficiency with the PL after the removal of the 'electrical 𝜂𝜂𝐵𝐵𝐵𝐵𝐵𝐵 = 5,47131 ∙ 10−17 ∙ 𝑃𝑃𝑃𝑃 + 0,940055947 ∙
outliers values' were processed and analysed, Fig. 8. The ∙ 𝜂𝜂𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 + 1,55431 ∙ 10−15 (4)
modelled global efficiency after the elimination of 'global
outliers values' is represented in the Fig. 9. Equations (2), An ANOVA analysis showed a poor significance of
(3) express mathematically the variation of the modelled the partial load coefficient, the 'p' value of significance
electrical and global efficiency, without outliers values, threshold greater than 0.05 (p = 0.4117). In the regression
with PL. model described by equation (7) we can not ‡š Ž—†‡–Š‡
2𝐸𝐸
’‘••‹„‹Ž‹–›‘ˆ–Š‡—ŽŽ ‘‡ˆˆ‹ ‹‡–‘ˆ’ƒ”–‹ƒŽ load. This is
𝜂𝜂𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 = 0.0054 ∙ 𝑃𝑃𝑃𝑃𝐶𝐶𝐶𝐶𝐶𝐶 + 0.4126 (2) a better understanding of the tabular values since January
2𝐸𝐸
𝜂𝜂𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 = 0,4372 ∙ 𝑃𝑃𝑃𝑃𝐶𝐶𝐶𝐶𝐶𝐶 + 0,4152 (3) 1st 2012 to January 14th 2012, when we have virtually the
same efficiency (80.22% - 80.28%) for totally different
The 0.54% slope of the regression line is observed in PL values (47.50% and 95.68%, non-functional M2). It
the variation of the modelled electrical efficiency. We can can be argued that the variable PL of the two-engine
say that, even in the case of the elimination of the outliers assembly, as considered, does not have a decisive
values from graph of the modelled electrical efficiency, significance in the multiple regression model created in
the PL has an insignificant influence. The ANOVA comparison with the modelled global efficiency output.
analysis on the two values ranges, the unbalanced model The coefficient of the modelled global efficiency's
of electrical efficiency and the partial load, reveals a 'p' variable and the intercept have values of very good
value of significance threshold higher than 0.05 (p = significance, p <0.05. Less than 1% of the modelled
0.4494), which leads to the assumption that the two value global efficiency variables are positioned outside the
ranges are not related, assuming the null hypothesis is confidence interval (95%).
true.
Analyzing the variance of the real efficiency in
relation to two variables: the PL and the modelled global

Fig. 8. Variation of modelled electrical efficiency without 'electrical outliers values' with partial load - normal operation mode.

7
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

Fig. 9. Variation of modelled global efficiency without 'global outliers values' with partial load - normal operation mode.

Fig. 10. Variation of real efficiency with modelled global efficiency and with partial load.

We would like to thank to Prof. Rodica Frunzulica and to Prof.

Florin Baltaretu for their help and support for this research.
6 Conclusions
References
The paper presents a mathematical model of calculation
of daily global efficiency in a cogeneration plant with two 1. University of Dundee under co-ordination of Dr. B.
reciprocating internal combustion engines. Previously Ramsay, Educogen, The European Educational
were obtained equations of interdependence between the Tool on Cogeneration, Second Edition, 8 (2001)
partial load of the thermal engines assembly and the http://citeseerx.ist.psu.edu/viewdoc/download?doi=
modelled thermal, electrical and global efficiency 10.1.1.618.8470&rep=rep1&type=pdf
(calculated with mentioned formulas from this
introduction). We obtained the equation of variation of the 2. G. Mărcuș, V. Iordache, R. Frunzulică, F.
real efficiency of CHP plant (the efficiency from the Iordache, RRIC, Efficiency analysis of a CHP
database) with the partial load and the equation of plant, based on reciprocating engines as prime
variation of the real efficiency of CHP plant (the movers and a hot water boiler as peak source, 9,
efficiency from the database) with the modelled No. 3, 278 (2018) http://rric.ro/revista.php?id=25
(calculated) efficiency. The values of the efficiency from 3. S. Sanaye, M. A. Meybodi, S. Shokrollahi,
the predictive model strongly determine the real Selecting the prime movers and nominal powers in
efficiency values from the database. The processing of the combined heat and power systems, Elsevier Ltd.,
database information has occasionally led to finding and Applied Thermal Engineering 28 (2008)
correcting a wrong recording data. The behaviour of the https://doi.org/10.1016/j.applthermaleng.2007.08.0
CHP plant has been observed by removing of the 03
dispersed values outside (the outliers values). The 4. ASHRE, ASHRAE Handbook - Heating,
variation trend of thermal and global efficiency increases Ventilating, and Air-Conditioning Systems and
with increasing of PL. The electrical efficiency does not Equipment, I-P Edition, (2008)
depend on PL variation. The obtained regression https://www.ashrae.org/technical-resources/ashrae-
equations were statistically analysed and conclusions handbook
were drawn on the meanings of terms, variables and 5. SR EN 15316-4-4, Energy performance of
coefficients. These equations provide the method to adjust buildings — Method for calculation of system
the efficiency calculated according to the mentioned energy requirements and system efficiencies, Part
formulas so that the final result to be as accurate as 4-4, (2017)
possible to the real conditions obtained in the operation of
the system. 6. SR CEN/TR 15316-6-7, Energy performance of
buildings - Method for calculation of system

8
E3S Web of Conferences 85, 01012 (2019) https://doi.org/10.1051/e3sconf/20198501012
EENVIRO 2018

energy requirements and system efficiencies, Part

6-7,(2017)