Bulletin of the Transilvania University of Braov Vol.

2 (51) - 2009
Series I: Engineering Sciences

THERMODYNAMIC OPTIMIZATION OF
SCREW COMPRESSORS

Camelia POPA1

Abstract: A suitable procedure for optimization of the screw compressor

shape, size, dimension and operating parameters is described here, which
results in the most appropriate design for a given compressor application
and fluid. It is based on a rack generation algorithm for rotor profile
combined with a numerical model of the compressor fluid flow and
thermodynamic processes. Some optimization issues of the rotor profile and
compressor parts are discussed, using 5/6 screw compressor rotors to
present the results. It is shown that the optimum rotor profile, compressor
speed, oil flow rate and temperature may significantly differ when
compressing different gases or vapors or if working at the oil-free or oil-
flooded mode of operation.

Key words: thermodynamics, optimization, screw compressor,

multivariable system.

1. Introduction specified in advance. Usually this is a

manual process where the designer makes
Screw compressors are therefore a prototype system which is tested and
efficient, compact, simple and reliable. modified until it is satisfactory. With the
They have largely replaced reciprocating help of a simulation model the prototyping
machines in industrial applications and in can be reduced to a minimum. Recent
refrigeration systems. advances in mathematical modeling and
Screw compressors can be either single computer simulation can be used to form a
or multistage machines. Multistage are powerful tool for the screw compressor
used for the compressor working with process analysis and design optimization.
higher pressure ratios, while the single Such models have evolved greatly during
stages are used either for low pressure oil- the past 10 years and, as they are better
free machines or moderate pressure oil- validated, their value as a design tool has
flooded compressors. A special challenge increased. Their use has led to a steady
is imposed upon the multistage compressor evolution in screw rotor profiles and
optimization, because not only the compressor shapes which should continue
compressor geometry parameters and in future to lead to further improvements in
operational conditions, but also the machine performance.
interstage pressures are optimized. A problem in optimization is a number
As other design processes, the design of of calculations which must be performed to
screw compressors is an interactive identify and reach an optimum. Another
feedback process where the performance problem is how to be certain that the
of the compressor is compared with those optimum calculated is the global optimum.

1
Dunrea de Jos University of Galai.
382 Bulletin of the Transilvania University of Braov Vol. 2 (51) 2009 Series I

Among the optimization methods In addition, the maximum delivery per unit
frequently used in engineering are steepest size or weight of the machine must be
descent, Newtons method, Davidson accompanied by minimum power
FletcherPowells method, random search, utilization for a compressor and maximum
grid search method, search along power output for an expander. This implies
coordinate axes, Powells method, Hooke that the efficiency of the energy
Reavess method. A widely used method interchange between the fluid and the
for optimization of functions with several machine is a maximum. Accordingly
optima is the genetic algorithm. It requires unavoidable losses such as fluid leakage
only a value of the target function and it and energy losses must be kept to a
can conveniently handle discontinuities, minimum. Therefore, increased leakage
however this method is slow in converging may be more than compensated by greater
to a solution. Alternatively a constrained bulk fluid flow rates. However,
simplex method, known as Box complex specification of the required compressor
method can be conveniently used. Box delivery rate requires simultaneous
complex method was therefore used here optimization of the rotor size and speed to
to find the local minima, which were input minimize the compressor weight while
to an expanding compressor database. This maximizing its efficiency. Finally, for oil-
finally served to estimate a global flooded compressors, the oil injection flow
minimum. That database may be used later rate, inlet temperature and position needs
in conjunction with other results to to be optimized. It follows that a
accelerate the minimization. The multivariable minimization procedure is
constrained simplex method emerged form needed for screw compressor design with
the evolutionary operation method which the optimum function criterion comprising
was introduced already in the 1950s by a weighted balance between compressor
[1,2]. The basic idea is to replace the static size and efficiency or specific power.
operation of a process by a continuous and
systematic scheme of slight perturbations 2. Minimization Method Used in Screw
in the control variables. Compressor Optimization
The effect of these perturbations is
evaluated and the process is shifted in the The power and capacity of contemporary
direction of improvement. The basic computers is only just sufficient to enable a full
simplex method was originally developed multivariable optimization of both the rotor
for evolutionary operation, but it was also profile and the whole compressor design to be
suitable for the constrained simplex performed simultaneously in one pass.
method. Its main advantage is that only a The optimization of a screw compressor
few starting trials are needed, and the design is generically described as a
simplex immediately moves away from multivariable constrained optimization
unsuitable trial conditions. problem. The task is to maximize a target
There are several criteria for screw function f (x1 , x 2 ,..., x n ) , subjected
profile optimization which are valid simultaneously to the effects of the explicit
irrespective of the machine type and duty. and implicit constraints and limits:
Thus, an efficient screw machine must
admit the highest possible fluid flow rates g i xi hi , i = 1, n
for a given machine rotor size and speed.
and
This implies that the fluid flow cross-
sectional area must be as large as possible. g i y i hi , i = 1, n + 1, m
Camelia POPA: Thermodynamic Optimization of Screw Compressors 383

respectively, where the implicit variables In the early 1960s, a method called the
y n+1 , y m are dependent functions of xi . simplex method emerged as an empirical
The constraints g i and hi are either method for optimization, this should not be
confused with the simplex method for
constants or functions of the variables xi . linear programming. The simplex method
When attempting to optimize a compressor was later extended by [3] to handle
design a criterion for a favorable result constrained problems. This constrained
must be decided, for instance the minimum simplex method was appropriately called
power consumption, or operation cost. the complex method, from constrained
However, the power consumption is simplex. Since then, several versions have
coupled to other requirements which been used. Here, the basic working idea is
should be satisfied, for example a low outlined for the complex method used. If
compressor price, or investment cost. The the nonlinear problem is to be solved, it is
problem becomes obvious if the necessary to use k points in a simplex,
requirement for low power consumption where k = 2n . These starting points are
conflicts with the requirement of low randomly generated so that both the
compressor price. For a designer, the implicit and explicit conditions in are
balance is often completed with sound
judgment. For an optimization program the satisfied. Let the points x h and x g be
balance must be expressed in numerical defined by
values. This is normally done with weights
on the different parts of the target function. ( ) ( ) ( ) ( )
f x h = max f x1 , f x 2 ,..., f x k
f (x ) = min f (x ), f (x ),..., f (x )
An important issue for real-world g 1 2 k
(1)
optimization problems is constraints. In the
general case, there are two types of
constraints, explicit or implicit. The 3. Calculation of Thermodynamic
explicit ones are limitations in the range of Processes in Screw Compressor
optimized parameters, for example Optimization.
available component sizes. These two
different constraints can, in theory, be The algorithm of the thermodynamic
handled more or less in the same way. In and flow processes used in optimization
practice, however, they are handled calculations is based on a mathematical
differently. model comprising a set of equations which
The implicit constraints are often more describe the physics of all the processes
difficult to manage than explicit within the screw compressor. The
constraints. The most convenient and most mathematical model gives an instantaneous
common way is to use penalty functions operating volume, which changes with
and thus incorporate the constraints in the rotation angle or time, together with the
objective function. Another way is to tell differential equations of conservation of
the optimization algorithm when the mass and energy flow through it, and a
evaluated point is invalid and generate a number of algebraic equations defining
new point according to some phenomena associated with the flow.
predetermined rule. Generally, it can be These are applied to each process that the
said that constraints, especially implicit fluid is subjected to within the machine;
constraints, make the optimization problem namely, suction, compression and
harder to solve, since it reduces the discharge. The set of differential equations
solution space. thus derived cannot be solved analytically
in closed form. In the past, various
384 Bulletin of the Transilvania University of Braov Vol. 2 (51) 2009 Series I

simplifications have been made to the cycle are derived from the internal energy
equations in order to expedite their and the volume and the computation is
numerical solution. The present model is carried out through several cycles until the
more comprehensive and it is possible to solution converges. The working fluid can
observe the consequences of neglecting be any gas or liquidgas mixture, i.e. any
some of the terms in the equations and to ideal or real gas or liquidgas mixture of
determine the validity of such assumptions. known properties. The model accounts for
This provision gives more generality to the heat transfer between the gas and
model and makes it suitable for compressor and for leakage through the
optimization applications. A feature of the clearances in any stage of the process. The
model is the use the energy equation in the model works independently of the
form which results in internal energy rather specification of compressor geometry.
than enthalpy as the derived variable. This Liquid can be injected during any of the
was found to be computationally more compressor process stages. The model also
convenient, especially when evaluating the takes in consideration the gas solubility in
properties of real fluids because their the injected fluid. The thermodynamic
temperature and pressure calculation is not equations of state and change of state of
explicit. However, since the internal the fluid and the constitutive relationships
energy can be expressed as a function of are included in the model. The following
the temperature and specific volume only, forms of the conservation equations have
pressure can be calculated subsequently been employed in the model. The
directly. All the remaining thermodynamic conservation of internal energy is:
and fluid properties within the machine

dU dV
= m& in hin m& out hout + Q& p (2)
d d

m& in hin = m& suc hsuc + m& l , g hl , g + m& oil hoil (3)

m& out hout = m& dis hdis + m& l ,l hl ,l (4)

The mass continuity equation is:

where h and x are angle and angular speed
of rotation of the male rotor respectively, dm
= m& in m& out (5)
h = h( ) is specific enthalpy, m& = m& ( ) is d
mass flow rate going in or out, p = p ( ) is
fluid pressure in the working chamber The instantaneous density = ( ) is
control volume, Q& = Q& ( ) , heat transfer obtained from the instantaneous mass m
between the fluid and the compressor trapped in the control volume and the size
of the corresponding instantaneous volume
surrounding, V& = V& ( ) local volume of the
V as = m V .
compressor working chamber. In the above
equation the index in denotes inflow and The suction and discharge port flows are
the index out the fluid outflow. Oil and defined by velocity through them and their
leakage are denoted by indices oil and l. cross section area
Camelia POPA: Thermodynamic Optimization of Screw Compressors 385

m& in = win in Ain , dToil hoil Aoil (T Toil )

(6) =
m& out = wout out Aout d m& oil coil
(8)
T kToil , p
The cross-section area A is obtained Toil =
from the compressor geometry and it was 1+ k
considered as a periodical function of the oil m& oil coil
k=
angle of rotation . hoil Aoil
Leakage in a screw machine forms a
substantial part of the total flow rate and k is, therefore, a time constant and h and
plays an important role because it affects A are the heat transfer coefficient between
the delivered mass flow rate and oil and gas and effective area surface based
compressor work and hence both the on the mean Sauter diameter d of the oil
compressor volumetric and adiabatic droplet. c is specific heat. is a time
efficiencies. step and index p denotes previous.
The solution of the equation set in the
form of internal energy U and mass m is
p 22 p12 performed numerically by means of the
m& l = wl l Al = (7)
p RungeKutta fourth order method, with
a + 2 ln 2
2
appropriate initial and boundary
p1
conditions. As the initial conditions were
arbitrary selected, the convergence of the
where a is the speed of sound, is a solution is achieved after the difference
compound resistance coefficient and between two consecutive compressor
indices l, 1 and 2 represent leakage, cycles becomes sufficiently small.
upstream and downstream conditions. Once solved, internal energy U ( ) and
Injection of oil or other liquids for mass in the compressor working chamber
lubrication, cooling or sealing purposes, m( ) serve to calculate the fluid pressure
modifies the thermodynamic process in a
and temperature.
screw compressor substantially. Special
effects, such as gas or its condensate Since U ( ) = (mu ) + (mu )oil , specific
mixing and dissolving in or flashing out of internal energy is:
the injected fluid must be accounted for
U (mcT )oil
separately if they are expected to affect the u= (9)
process. In addition to lubrication, the m
major purpose for injecting oil into a As volume V ( ) is known, a specific
compressor is to seal the gaps and cool the
gas. Flow of the injected oil, oil inlet volume is calculated as v = V m .
temperature and injection position are Therefore, temperature T and pressure p
additional optimization variables if the oil- for ideal gas can be calculated as:
flooded compressors are in question. Heat
u RT
transfer between oil and gas is modeled as T = ( 1) , p= (10)
a first order dynamic system. R v
386 Bulletin of the Transilvania University of Braov Vol. 2 (51) 2009 Series I

where R and are gas constant and adiabatic. Isothermal Wt and adiabatic
isentropic exponent respectively. In the work Wa and are given here for ideal gas.
case of a real gas, u = f1 (T , v ) and
p = f 2 (T , v ) are known functions and
should be solved to obtain the fluid
m = min mout , Wind = Vdp
cycle
(12)

temperature and pressure T and p . This Wind z l n

task is simplified because internal energy m& = mz l n 60 , Pind = (13)
u is not a function of pressure, therefore,
60
f1 and f 2 can be solved in a sequence. In
(F1n )
+ F2 n Lnz1
m& s = (14)
the case of a wet vapor because of the fluid 60
phase change either through evaporation or p
condensation, the saturation temperature W1 = RT1 ln 2 ,
p1
and pressure determine each other between (15)
themselves and also the liquid and vapor
W2 = R(T2 T1 )
internal energy and volume, u and v . 1
Indices f and g denote liquid and gas
phases. Therefore, vapor quality x can be A full and detailed description of the
calculated by successive approximations of presented model of the compressor
u . Variables T or p and v can be thermodynamics is given in [7].
obtained from: m& W
v = , t = t ,
m& s Wind
u = (1 x )u f + xu g , (16)
(11) W P
v = (1 x )v f + xv g i = a , Ps = ind
Wind V&
Numerical solution of the mathematical Compressor speed is used as the
model of the physical process in the compressor operating variable and oil
compressor provides a basis for a more flow, temperature and oil injection position
exact computation of all desired integral are oil optimization parameters. Each of
characteristics with a satisfactory degree of these rotor variables has its own influence
accuracy. The most important of these upon the compressor process which is
properties are the compressor mass flow qualitatively explained in the following
rate m& [kg/s], the indicated work Wind qualitative diagrams, Figs. 1 and 2.
[kJ] and power Pind [kW], specific Compressor shaft speed increases
dynamic losses and decreases relative
indicated power Ps [kJ/kg], volumetric leakages. These two opposing effects cause
efficiency v , adiabatic efficiency i and that therefore, an optimum value of the
isothermal efficiency t . Z1 and n are the shaft speed exists which gives the best
number of lobes in the main rotor and main compressor performance.
rotor rotational speed. F1 , F2 and L are In oil-flooded compressors oil is used to
lubricate the rotors, seal the leakage gaps
the main and gate rotor cross section and and cool the gas compressed. Therefore its
length. Index s means theoretical and influence upon the compressor process is
indices t and a denote isothermal and complex.
Camelia POPA: Thermodynamic Optimization of Screw Compressors 387

Dynamic Friction
drag
Loss
Loss

Leakage
Volumetric

Oil flow
Shaft speed

Fig. 1. Influence of the compressor shaft Fig. 2. Influence of the injected oil flow
speed upon the compressor volumetric upon leakage and friction drag losses
and dynamic losses

More oil improves the compressor Computation of the instantaneous cross-

volumetric efficiency and also improves sectional area and working volume could
cooling, however, it increase the friction thereby be calculated repetitively in terms
drag between the rotors themselves and of the rotation angle. A mathematical
between the rotors and housing. Obviously model of the thermodynamic and fluid
an oil flow rate exists which will produce flow process is contained in the package,
the best compressor performance. Each of as well as models of associated processes
the described geometry and operating encountered in real machines, such as
parameters influences the compressor variable fluid leakages, oil flooding or
process on its own way and only a other fluid injection, heat losses to the
simultaneous minimization, which takes surroundings, friction losses and other
into consideration all the influences effects. All these are expressed in
together will produce the best overall differential form in terms of an increment
compressor performance. Therefore only a of the rotation angle. Numerical solution of
multivariable optimization finds its full these equations enables the screw
sense in the evaluation of the best compressor flow, power and specific
compressor performance. power and compressor efficiencies to be
calculated.
4. Conclusions
References
A full multivariable optimization of
screw compressor geometry and operating 1. Box, M.J.: A new method of
conditions has been performed to establish constrained optimization and a
the most efficient compressor design for comparison with other methods. In:
any given duty. This has been achieved Computer Journal 8 (1965) 4252.
with a computer package for modeling 2. Fleming, J.S., Tang, Y., Xing, Z.W.,
compressor processes, developed by the Cook, G.: The use of superfeed in a
authors, which provides the general refrigeration plant with a twin screw
specification of the lobe segments in terms compressor: an optimization technique
of several key parameters and which can for plant design. In: IIR XIX
generate various lobe shapes and Congress, The Hague, The
simultaneously calculates compressor Netherlands, 1995.
thermodynamics.
388 Bulletin of the Transilvania University of Braov Vol. 2 (51) 2009 Series I

3. Hanjalic, K., Stosic, N.: Development Schraubenmaschinen 94, Dortmund

and optimization of screw machines VDI Berichte 1135, 1994.
with a simulation model, Part II: 6. Stosic, N., Hanjalic, K.: Development
Thermodynamic performance and optimization of screw machines
simulation and design. In: ASME with a simulation model. Part I: Profile
Transactions, Journal of Fluids generation, In: ASME Transactions,
Engineering 119 (1997) 664. Journal of Fluids Engineering 119
4. Hanjalic, K., Stosic, N.: Application of (1997) 659.
mathematical modeling of screw 7. Stosic, N., I.K., Kovacevic, A.:
engines to the optimisation of lobe Optimization of screw compressor
profiles. In: Proc. VDI Tagung design. In: XVI International
Schraubenmaschinen 94 Dortmund Compressor Engineering Conference
VDI Berichte Nr. 1135, 1994. at Purdue, July, 2002.
5. Sauls, J.: The influence of leakage on
the performance of refrigerant screw
compressors. In: Proc. VDI Tagung