ESL-IE-85-05-01

SMALL-SCALE INDUSTRIAL COGENERATION: DESIGN USING RECIPROCATING

ENGINES AND ABSORPTION CHILLER

Joseph R. Wagner

Mechanical Technology Incorporated

Latham, New York

ABSTRACT
This paper describes a packaged cogeneration system
designed for light industrial applications (i.e.,
situations where a user wants a maximum of 1 MW of
cogenerated electricity). The design employs reci
procating engines fueled with natural gas or liquid
fuels. Waste heat from the engine exhaust and jacket
water is used to drive an absorption chiller. This
approach yields a variety of unique advantages,
including: (1) ability to satisfy the needs of facil
ities that have a low ratio of thermal-to-electric
energy needs, and which have significant. refriger
ation loads, and (2) the ability to provide greater
operating efficiency in the <1 MW size range, espe
cially at part load, when compared to systems based
on gas turbines or steam turbines.
Although component substitution can be used to vary
the energy output capabilities of the system, the
discussion focuses' on a unit having an electrical
rating of 418 kW and cooling capacity of 160 tons.
This paper discusses component selection, operating
parameters, economics, maintenance issues, and other
factors.
INTRODUCTION
Over the last dozen years there has been a surge of
interest in industrial cogeneration systems (i.e.,
systems which simultaneously produce heat and elec
tricity). This surge began with the 1973-74 Energy
Crisis, and was in response to dramatic increases in
the cost of fossil fuels. Despite the stabilization
of fossil fuel prices in the 1980s, interest in
cogeneration has continued to grow. This can be
credited to a number of factors, including the demon
strated success of cogeneration projects installed
during the last decade and the increasing availabili
ty of proven equipment to meet the needs of different
types of industrial facilities.
Despite the increasing variety of options, there are
still applications where suitable equipment is diffi
cult to find. Small-scale cogeneration is one such
application.
For the purpose of this paper, small
scale means that the user wants a continuous output
of, at most, 1 MW of cogenerated electricity. The
plant's total electrical demand might in fact be much
larger.
There are three basic strategies for cogeneration,
categorized by the primary fuel-using device. These
are:
Gas turbine
Steam boiler
Reciprocating engine.

Plant operators often find gas turbines to be unsuit

able for small-scale applications. One important
factor in this determination is the generally: low
fuel-to-shaft efficiency of gas turbines in this,size
range. The percentage of the fuel energy that is
converted to electricity is significantly lower .than
is typical of larger gas turbines and this, in turn,
can undermine project economics. A steam bQiler
system, with a backpressure turbine for electrical
generation, likewise generally converts a relatively
small portion of the fuel energy into electri,ity.
This is especially true if the boiler operates at the
modest pressure levels typical of small plants. '
For reasons such as these, it is known that the first
two strategies above do not meet the requirements of
all plants.
This paper proposes an alternative
design based on the third strategy, namely, recipro
cating engines.
.
PROS AND CONS OF RECIPROCATING ENGINES
Relative to the prime movers used in other cogenera
tion strategies, reciprocating engines tend to offer
the following advantages:
Low capital cost, on a dollar per shaft ~orsepower basis
High fuel-to-shaft efficiency
Good part-load efficiency
Good load following
No need for gas compressor (as is the; case
with gas turbines) when fueled with naltural
gas
Speed range allows direct drive of gener~tors
Many suppliers and wide selection in th~ <1MW
size range.
Disadvantages include relatively high maint~nance
costs, on a C/kWh basis. This tends to be espedially
true if the engine is fueled with natural gas. : Such
engines generally were designed initially to: burn
diesel fuel (i.e., distillate oiU, and have' been
converted to natural gas by the addition of :spark
plugs and other modifications.
Depending on the
quality of the modifications, problems can aris~ in a
number of areas, including the spark plug$ and
7xhaust valves. Aside from the expense of maiqtain
Ing these components, unscheduled outages can:occur
and, in some cases, can severely affect p~oject
economics.
Another disadvantage of reciprocating engines i$ that
they are not good steam generators, at least when
compared to gas turbines. Vi rtuall y all the :waste
heat leaving a gas turbine is in the form of
high-temperature gas, and can be readily conver~ed to
process steam at pressures typical of the main:steam

Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985

ESL-IE-85-05-01

header at a small plant (e.g., 125 psig). Recipro

cating engines, on the other hand, reject about half
their waste heat to the cooling water circulating
through their engine blocks. This so-called jacket
water exi ts the block at only about 200 0 F. Conse
quently, the most successful cogeneration applica
tions for these engines typically involve situations
where the user has a direct need for hot water.

With this approach, a number of grid interconnect

schemes are possible. Direct connection to the grid
could be maintained at all times, allowing the plant
to buy or sell power. Alternatively, the grid could
be used only for backup. That is, the cogeneration
system would supply all plant power. If the cogener
ation system failed, or could not satisfy the load,
the system would shut down and the plant would tempo
rarily reconnect to the grid to buy power. Another
approach is the stand-alone option. This requires
synchronous generators, a degree of overcapacity or
loadshedding ability, and generally a high level of
redundancy throughout the cogeneration system.

SYSTEM DESIGN APPROACH

Choice of fuel is the first issue confronted by an
aspiring cogenerator. The motivation to cogenerate
is a function of the difference between the price of
purchased electricity and the price of the fuel used
by the cogeneration system. The greater this differ
ence, the greater the savings potential. Generally
speaking, this dictates using the cheapest fuel, on a
$/MMBtu basis, that the system can reasonably accom
modate without unduly increasing capital or mainte
nance costs.

The local ut i li ty' s rate schedule, interconnect ion

specifications, and general attitude towards cogener
ation normally dictate which approach is appropriate.
The present system is designed for direct grid
connection, and economic calculations are based on
the assumption that the plant's base electricity
consumption exceeds coge'neration capacity, such that
the plant would always be buying power from the grid,
and the cogeneration system would always be heavily
loaded. This situation is economically equivalent to
a case where the cogenerated electricity exceeds
plant demand, and the utility buys the excess at a
price equal to the utility's selling price.

For reciprocating engines in the United States, this

often leads to the conclusion that natural gas is the
fuel of preference. As compared to the diesel engine
from which it is derived, a natural gas engine will
typically be derated, somewhat less efficient, more
expensive to maintain, and more expensive on a $/kW
basis. Yet these disadvantages are frequently more
than offset by the low price of natural gas relative
to diesel fuel. This is especially true where the gas
utility is providing preferential rates, or where
users produce their own gas. Since small-scale
cogeneration is a developing market, these high
potential applications should be addressed first.
Thus, it was decided to focus the present design
effort on engines fueled with natural gas.

Another decision that must be made regarding basic

design has to do wi th the engine's waste heat. As
noted above, a reciprocating engine is economically
most attractive in situations where the hot jacket
water can be used directly. Unfortunately, most
plants have limited use for any significant new
source of hot water. Applications such as domestic
hot water, space heating, and boiler water preheating
are usually small, seasonal, or are easily satisfied
by recovering waste heat from existing plant equip
ment. Potentially, the jacket water could be sent to
an absorption chiller, but the coefficient: of
performance (COp) for such units is low, andl the
resulting refrigeration tonnage is small.

Aside from fuel type, choices must be made regarding

other design options, as indicated in Figure 1. The
shaft power can be used to drive either a
variable-speed electric generator, which would
require a dedicated circuit to a variable frequency
load, or a fixed-speed generator, or to drive mechan
ical loads such as compressors, pumps, and fans. The
present system was designed to drive a fixed-speed
electric generator, since 60-Hz electricity can be
used directly throughout a plant, with a minimum of
site engineering. This makes the system easier to
retrofit to a large population of plants.

Heat Utilization

Electric
Connection
and Fuel

Thi s often leaves the engine's exhaust gas as the

only useable form of heat. It can be used to generate
a small amount of steam, or possibly can be linked to
a desiccant dehumidifier to help reduce the load on
the plant's refrigeration and air conditioning
system. Another option is to send the exhaust gas to
an absorpt ion chi ller, where it can be used to
produce chilled water. The COP is typically signif
icantly greater than for a chiller driven by hot
water. The chilled water can be used directly for
refrigeration, air conditioning, or process cooling,
or could be used as part of a refrigeration cascade,
as will be described.

Electromechanical
Drives
Direct Drive of
Compressors, Pumps, etc.

Hot Water
Space Heating

The chiller option was the one selected for the pres
ent design, because it was judged to be applicable to
the largest number of plants. A contributing factor
to this decision was the identification of a chiller
that could simultaneously use both the exhaust gas
and the jacket water. Thus, this strategy allows use
of all the engine's waste heat and converts it into a
form deemed most likely to be useable at a typical
plant.

Generate Variable
Frequency Electricity

Dehumidification
Rerrigeration
- Direct
- Cascade

Generate 60 Hz
Electricity
Diesel
Other

To further increase the system's appl icabi 1 i ty, it

was decided to design it as a packaged module. This

8:)4242-1

Figure 1.

System Design Options

Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985

ESL-IE-85-05-01

or is shut off for routine serv1C1ng, the util!ity

sees this as a large demand spike. Depending on .the
local rate schedule, this can have a signifidant
effect on project economics, since demand chargeS in
some service territories are very high. Another
advantage of the mult~ngine approach is that it is
more adaptable to stand-alone strategies, where
redundancy and ability to modulate system output 'are
more important.

reduces site engineering and installation costs. It

also allows better integration of the system compo
nents. The final system is designed to be a factory
assembled, standardized product capable of achieving
economies accruing from volume production.
System Layout
~igure

2 is a schematic showing the functional

relationship between basic system components. The
system shown contains three engines, each of which
drives an electric generator. Multiple engines were
specified to avoid creating demand peaks. A single,
larger engine-generator can do the work of three
smaller units, but when a single, large engine fails

The actual physical dimensions of the proposed system

are indicated in Figure 3. Thi s layout depicts a
three-engine skid that can be transported on a stan
dard 45-ft trailer. Gross vehicle weight limitations
would probably require that the engine-generators be
transported separately. The engine generators are
designed for easy changeout, primarily to satisfy
servicing requirements.
This feature also allows
simplified installation.
'

Electricity

to Building

Electricity
from Grid

The total basic continuous-duty electric generating

capacity for all three engines is approximately, 418
kW. Depending on the components selected, the b~sic
layout's electric generation capacity can be designed
to be as low as 279 kW or as high as 626 kW. The high
er capacity figure would require turbocharged
engines. Capacities of 1 MW or more could be achieved
by selecting engines having larger displacements, but
this would require at least two skids, and would tend
to violate the concept of a packaged system.
.

Switchgear I Controls

Engine-Generators

----- - --- ------,I

I
I

I
:
I
1

1
I

Cooling Water

Circulated to

Cooling Tower

~t:[=]--C2SZ:::}-~I~~EXhaust
J
to Stack
Chilled Water
to Building

44 F

Exhaust Gases
1200F

En ine Jacket Water

Refrigeration Cascade
As shown 1n Figure 2, the system is des igne~ to
produce chi lled water at about 44 0 ~. Thi s is: the
approximate lower limit of the refrigeration te~per
at~re achievable with a lithium-bromide absorp:tion
ch1ller. Present ly, th i sis the only chi ller t.ech
nology identified as commercially available and c~pa
ble of satisfying various technical constraints.
This would seem to limit application of this system
to plants having refrigeration loads of 44~ and
hi~her. In fact, this is not so, since a refri1ger
at10n cascade can be employed.

54F

Chilled Water
Return

200F

Jacket Water Return

to Engines

Space Heat
I1::;::-Cooling Water
I -. for Trim

I ...

r-->-tl----------
Domestic
Hot Water

City Water
Figure 2.

Figures 4 and 5 provide a simple example of a re~rig

eration cascade applied as a retrofit. ~igu~e 4

a33S762

Cogeneration System Schematic

Engine-Generator Sets

Electric
Switch
Gear

.~:::o~.

1:,:

u'
~l-~--~~--~~~-~_-~L----------;-;-;-o-;-;-n-~-;-u-;-p--~~~~~~~~========='~J _~
Water Pump
Cooling

-~-~--==.&:-------:.::~:~--

Absorption
Chiller

12 It i

'

'---------------45 ft----------------------
Figure 3.

Plan View: Equipment Layout

3
Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985

ESL-IE-85-05-01

shows a plant refrigeration system prior to retrofit.

Refrigerant R-12 (a type of Freon"') is arbitrarily
assumed as the working fluid in a system maintaining
various refrigeration loads at OOF (i.e., OOF suction
temperature).
A compressor takes in the 24 ps ia
Freon vapor exi t ing the heat exchanger( s) in the
refrigerated space(s), and compresses this vapor to
150 psia.

The overall system is normally set up so that the

expansion valves have a relatively uniform upstream
pressure. This requires that the system be set up for
"worst day" conditions, and results in the COmprE!S
sors having to contend with a relatively high exhaust
pressure, regardless of weather conditions. This
exhaust pressure is the pressure that the condensers
can ensure will not be exceeded, unless weather
conditions exceed the "worst day" design conditions.
As shown in the example, the worst day is anticipated
to be one where the condensing temperature would be
110F, with the compressors working at full capacity.
This is a typical design temperature in the U.S.
Condenser sections are turned on or off to maintain
this temperature, and hence to maintain the fixed
pressure of 150 psia.

This exhaust pressure is a function of the operation

of the condensers employed in the refrigeration
system. The performance of the condensers is, in
turn, governed by ambient weather conditions. On a
hot day the condensers must work harder to extract
the necessary amount of heat from the Freon vapor, so
it will condense and allow the refrigeration cycle to
function.
If condensation does not occur fast
enough, pressure increases on the condenser side of
the system.

Figure 5 shows how this situation would change by

instituting a refrigeration cascade. A new heat
exchanger would be installed to allow condensation to
be performed by 44F chilled water, as supplied by a
lithium-bromide chiller. The chilled water temper
ature is not subject to ambient conditions, and would
allow readjustment of the refrigeration system. The
example shows this readjustment resulting in a new
condensing design temperature of 60F, which reduces
the controlled pressure on the condenser side of the
system to 72 psia. This, in turn, requires less work
from the compressors and results in lower power
consumption.

Staged Condensers

,--------------------,

:J,
I
I

I
I

I
I

Heat Reclaim

I
IL

for DomestIc ____

HoI Waler and
Space Heat

Expansion

180'F 150psia

Table 1 illustrates the power savings achievable by

reducing the condensing temperature to 60F versus an
original 10aoF. The savings are stated in terms of
compressor horsepower per ton of refrigeration (Note:
1 ton = 12,000 Btu/h). The savings increase signif
icantly as the refrigeration temperature decreases.
The far right-hand column in Table 1 shows the
required refrigeration tonnage that must be supplied
by the absorption chiller, per ton of total refriger
ation load, in order for the stated compressor
savings to occur. These data imply that the chilled
water is increasingly effective when used in a
cascade maintaining lower refrigeration temper
atures.

150p5ia 110'F

Valve

O'F
24 psia
Receiver
Rerrigeration

Loads

Compressor

Figure 4.

Existing Refrigeration System

(R-12 V~lues Used for Illustration)

Staged Condensers

Table 1. Approximate Benefit of Using Cogener-

ated Chilled Water to Reduce Refriger
ation Loads

,--------------------,
Heat
Exchanger

:=
I

1
I

Evaporator
Temp. (OF)

I
I
I

Heat Reclaim
for Domestic - 4

Hot Water and

Space Heat'

I
L

-----
Expansion
Valve

120'F 72 psia

72 psia 60'F

Receiver
Compressor

Refrigeration
Loads

Addllional domestic hoi water and space heal now provided by engine waste heat
a:J..tCle71

Figure 5. Refrigeration System After Retrofit

(R-12 Values Used for Illustration)

Chiller
Tonnage/Ton
Savings Refrigeration

108F

60F

40

1.27

0.33

0.94

1.07

30

1.54

0.51

1.03

1.11

20

1.84

0.71

1.13

1.15

10

2.17

0.93

1.24

1.20

2.55

1.18

1.37

1.25

-10

2.98

1.45

1.53

1.31

-20

3.48

1.74

1.74

1.37

-30

4.05

2.08

1.97

1.44

-40

4.73

2.45

2.28

1.52

O'F
24 psia

Horsepower/Ton va
Condensing Temperature

"Suction temperature

Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985

834176

ESL-IE-85-05-01

Table 3. Thermal Energy Characteristics of C~ins

G-8SS Gas Engine (12:1 Compression ~atio,
1800 rpm)

Component Selection
A variety of reciprocating engines is available for
incorporation into the basic system design. Table 2
shows some examples of commercially available
engines, along with approximate technical character
istics. For illustration purposes, the Cummins G-855
is assumed to be the prime mover for the basic system
described in this paper. It has been selected pri
marily because it exhibits high efficiency relative
to other engines in the approximate 200-hp size
class. The 200-hp size class was selected as being
consistent with the multiengine strategy and with the
desired overall electrical and thermal output of the
system.

Energy
Split

Equivalent
Tons of
Cooling

31.1

100% Load

450.0

21.4

75% Load

390.4

18.5

50% Load

314.6

14.9

100% Load

358.3

32.3

75% Load

293.8

2~.5

50% Load

242.8

21:.9

Mechanical Efficiency (%)

Jacket Water (10 3 Btu /h)

Table 3 provides additional detail on the mechanical

and thermal output of the Cummins G-855. The column
on the right side of Table 3 indicates the tons of
refrigeration
that
could
be
produced
by
a
lithium-bromide chiller being driven by the engine's
waste heat. Note that these values are for only one
engine. The design specifies that the output of
multiple engines will go to a single chiller. There
fore the values in Table 3 must be multiplied by the
number of engines to show total system performance.

Exhaust* (10 3 Btu/h)

The chilled water tonnage values are based on the

performance characteristics of a Hitachi Paraflow"
chiller. To date, this is the only commercially
available chiller in the U.S. which has been identi
fied for this application. This unit can use engine
jacket water and exhaust gas simultaneously, and
overall can provide an estimated 53.7 tons of refrig
eration using the waste heat from one Cummins G-855.
It has been suggested that the smallest Hitachi unit
is, in fact, oversized for this application, but this
does not affect performance estimates.

Total Tons of Cooling

Generated by Chiller
at 100% Load

i
53.7
I

852193

*Based upon reducing exhaust temperature to 375P

I
!
i

though there is more waste heat in the jacked water

than in the exhaust gas. The reason for this is that
the chiller COP increases with increasing :engine
waste heat temperature, and the exhaust gas is
significantly hotter than the jacket water.

The chiller produces significantly more tonnage from

the exhaust gas than from the jacket water. This is
shown graphically in Figure 6. Thi s occurs even

ECONOMICS
Table 2.

Comparison of Gas Engine Efficiencies

(Engines in Nominal 200-hp Class)

Parameter

Cummins
G855

Caterpillar
3308

Waukesha
F1905GR

Displacement (In.3)

855

638

638

1905

Aspiration

Max. Continuous-Duty
Horsepower

200

145

220

240

1800

1800

1800

1200

100% Load

0.311

0.297

0.291

0.259

75% Load

0.288

0.280

0.278

0.242

50% Load

0.250

0.246

0.247

0.207

Speed (rpm)

To determine the economic viability of the p~oposed

system, an estimate of potential user benefi'ts was
developed. The benefits were calculated under the
following economic assumptions:
* System capacity: 418-kW electricity; l~O-tons
chilled water
,
* All electricity produced is used on sit~
No stand-alone capability
i
80% load factor (average = 334 kW, 129 ~ons)
Maintenance: ie/kWh produced ($29,000/Ytr)
Average fuel-to-electricity efficiency: 28.8%
Average chiller COP: 0.80 (combined ~xhaust
and jacket water).

Fuel-lo-Shaft
Efficiency:

*N = Natural: T = Turbocharged
**On a higher heating value basis (1028 Btu/sci)

Once these assumptions are fixed, user b~nefits

become a function of energy prices. Using a:2-year
simple payback as a "hurdle rate," the minimu, level
of annual user benefits required to make the'system
economically attractive is equal to the installed
cost divided by two. Conversely, the maximum allow
able installed cost that a user can tolerate is equal
to estimated annual savings times two.

834177

5
Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985

ESL-IE-85-05-01

60
a>
c::
'0,

Superimposed on this figure are lines showing antlcl

pated installed costs. The "retrofit" line shows the
estimated installed cost if the system is assumed to
be equipped with its own enclosure and to bear the
full cost of all interconnections with existing plant
systems. The "new construction" line shows the esti
mated installed cost assuming that the system is
incorporated as part of a new plant, or as part of a
plant expansion or modernization, in which case many
cost savings can be realized in the areas of enclo
sures, piping, controls, and heat exchangers.

50

c::
w
......
c::
0

+::

III
~

a>

Ol
"C

'$
II:

The result is that even in the "new construction"

scenario, system economics will likely be signifi
cantly attractive only in those areas of the country
having relat i vely high elect ric rates and low-to
moderate natural gas rates. Nonetheless, these areas
exist and are large enough to constitute a signif
icant market for the system. Candidates include
southern California, New York City, Long Island, and
Connecticut, where there are favorable energy prices,
large populations, and many light industrial facili
ties. These areas also contain many large commercial.
facilities, e.g., retail establishments and apartment
complexes, that can employ this system.

'0
III

c::

20
Jacket Water
Tonnage

10

'--_--JI-.._.......L_ _.......L_ _--L.._ _.....I

50

60

70

80

90

% of Rated Engine Load

100
8521114

Figure 6. Absorption Chiller Performance Using

Cummins G-855 Gas Engine (12:1
Compression Ratio, 1800 rpm)

FUTURE REQUIREMENTS
In order for the proposed system to appeal to a wider
market, i.t is clear that user payback must bE~
enhanced. Fortunately, many avenues exist for bring-'
ing this about. A full discussion of these avenues is
beyond the scope of this paper, but a brief listing in
in order:

This principle, which presumes the user requires a

simple payback of 2 years or less, has been used to
generate Figure 7. For example, Figure 7 shows that
if electricity costs 8e/kWh and natural gas is avail
able at $4/MMBtu, the user can spend at most about
$310,000 and still achieve a 2-year payback on the
system. The $310,000 includes all equipment and
installation costs.

Engine Improvements
optimization for natural gas, less reliance
on off-the-shelf diesel hardware
- adiabatic components (i.e., insulate and
temperature-toughen engine so that more of
the waste heat goes into exhaust, thereby
improving chiller COp)

Price of Natural Gas ~

($/MMBtu)
4

500
-"
(.l
III
.0

>

Chiller Improvements
- optimize $/ton for this size range

III

Q.

... >-

Retrofit*

400

N~
o~

Better Component Integration.

.... '0

~o 300

SUMMARY

00
'0
'Oc;
Gllll

Many types of cogeneration systems have been intro-

duced and proven successful in recent years. None
theless, small-scale cogeneration still presents
special problems. This paper has described a system
design that may be appropriate where other approacheE;
are unattrac'tive.
The system uses reciprocating
engines, and typically is fueled with natural gas. A
typical configuration would be rated at 418 kW conti-
nuous-duty electrical output, but capacity can be
varied over a range of 279 kW to 626 kW by making
simple component substitutions.

=0

Ill:;,

<;;0
c;S::

-~

200

CIl

:0
III
~

.2
C(

100

Electricity Price (/kWh)

The engine waste heat is used to drive an absorption

chiller. The typical configuration would be rated at
roughly 160 tons refrigeration capacity. This capac-'
ity can be achieved by using the chiller output
directly, or by employing a refrigeration cascade to
carry low-temperature refrigeration loads.

85219t1i

*100th Unit Estimates

Figure 7.

Allowable Installed Cost, as Function

of Energy Prices

6
Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985

'.,
ESL-IE-85-05-01

sign i ficant market.

More importantly, poten ial
system enhancements have been identified and c uld
significantly lmprove user economics, such that the
system could appeal to a much larger market.

Economic projections indicate that presently the

system is attractive only in selected geographic
areas. Nonetheless, these areas have large populations of potential users and could provide a

Proceedings from the Seventh National Industrial Energy Technology Conference, Houston, TX, May 12-15, 1985