Description:5 PREAMBLE OF THE DESCRIPTION:
THE FOLLOWING SPECIFICATION PARTICULARLY DESCRIBES THE INVENTION AND THE
MANNER IN WHICH IT IS PERFORMED.
A. TECHNICAL FIELD OF THE INVENTION
[001] The present invention relates to the design, optimization, and operation of multi10 stage screw compressors. The present invention more particularly relates to a improving
mechanical efficiency in the multi-stage screw compressors in real-time, adaptive
modelling framework integrating Machine Learning (ML) and Bayesian Optimization (BO)
using a customized control unit.
15 B. BACKGROUND OF THE INVENTION
[002] Screw compressors are widely used in industries like manufacturing, chemical
processing, and oil & gas, where they run continuously under high pressure. However,
current methods for designing and optimizing these compressors come with several
challenges, making them inefficient and costly to operate. Traditionally, engineers
20 manually adjust key settings like interstage pressure and rotor speed using trial-and-error
techniques and historical data. This process is slow, prone to mistakes, and doesn’t adapt
well to changing conditions, often leading to poor performance. Because these systems
rely on static models, they require frequent recalibration, making it hard to maintain
efficiency over time.
25
[003] One major problem is optimizing key parameters like interstage pressures and
volume ratios. Current approaches involve guesswork and manual tuning, which can
result in excessive energy consumption. If these settings are even slightly off, the
compressor can use more power than necessary, increasing operational costs. Since
30 specific power consumption (SPC) is a crucial measure of efficiency, optimizing it requires
fine-tuning multiple interconnected parameters, which is difficult to achieve using
traditional methods. Additionally, most compressor models are static, meaning they don’t
automatically adjust to changes in conditions like suction pressure, temperature, or gas
composition. This makes them less flexible and unable to operate efficiently in real-world
35 industrial environments where conditions are constantly shifting.
[004] Several software tools, such as SCORG and CFD (Computational Fluid Dynamics)
simulations, are commonly used to model and predict compressor performance.
However, these tools don’t provide real-time optimization. Engineers must manually
2
5 input and adjust parameters, making the process time-consuming and expensive. Some
companies use optimization algorithms like Genetic Algorithms (GA) or Particle Swarm
Optimization (PSO) to improve performance, but these methods don’t continuously adapt
to real-time data. They work well in predefined scenarios but lack the ability to make
ongoing adjustments, especially in multi-stage compressors where different stages must
10 work together efficiently.
[005] Many industrial screw compressors also rely on rule-based control systems that
maintain basic settings like pressure and flow rates. However, these systems follow
predefined rules and don’t optimize performance dynamically. They don’t use predictive
15 analytics or advanced optimization techniques, meaning they miss opportunities to
improve efficiency and reduce operating costs.
[006] To address these issues, the present invention introduces an automated, real-time
optimization system that integrates Machine Learning (ML) and Bayesian Optimization
20 (BO) to improve screw compressor performance. Instead of relying on manual tuning, the
system continuously analyzes historical and real-time operational data using a machine
learning model. It then applies Bayesian Optimization to fine-tune parameters like
interstage pressure, rotor speed, and volume ratios, ensuring that the compressor runs at
peak efficiency with minimal human intervention. Because the system learns and adapts
25 over time, it automatically adjusts to changing conditions, eliminating the inefficiencies of
traditional models.
[007] This new approach offers major benefits over existing methods. It removes the need
for trial-and-error adjustments, reduces energy consumption, and improves overall
30 efficiency. The system dynamically updates performance settings, leading to lower
operational costs and reduced wear and tear on the compressor. Additionally, it is scalable
and can be applied to different compressor sizes and configurations, making it a gamechanging advancement for industries that rely on screw compressors. By addressing the
weaknesses of traditional methods, this invention introduces a smart, self-improving
35 system that keeps compressors running efficiently, saving both energy and money while
enhancing reliability and performance.
C. OBJECT OF THE INVENTION
[008] The primary object of the invention is to develop an advanced modelling framework
40 that utilizes Machine Learning (ML) and Bayesian Optimization (BO) to optimize the
3
5 interstage pressure and key performance parameters of multi-stage screw compressors.
Specifically, this invention enhances the efficiency of a two-stage screw compressor
ensuring precise interstage pressure control to improve compressor performance and
energy efficiency.
[009] Another object of the present invention is to replace conventional trial-and-error
10 methods and static simulation models with advanced data-driven optimization. By
integrating Gaussian Process Regression (GPR) and Bayesian Optimization, the system
efficiently determines optimal compressor parameters, reducing manual intervention and
enhancing accuracy in achieving desired pressure ratios, flow rates, and energy
consumption.
15 [010] Yet another object of the present invention is to enable real-time adjustments of
compressor parameters based on operating conditions such as load, temperature, and
pressure variations. Unlike existing static models, this invention ensures continuous
optimization during operation, leading to sustained peak efficiency without requiring
manual recalibration.
20 [011] Yet another object of the present invention is to overcome the efficiency loss
experienced by multi-stage screw compressors operating at higher pressure ratios. By
dynamically optimizing interstage pressures using Bayesian Optimization techniques, the
system minimizes energy consumption while maintaining high performance, reducing
operational costs significantly.
25 [012] Yet another object of the present invention is to achieve performance
improvements without requiring expensive hardware modifications.
[013] Yet another object of the present invention is to allow predictive analytics to identify
performance inefficiencies before they occur. The GPR model anticipates potential
operational issues, while Bayesian Optimization continuously refines system parameters
30 to prevent inefficiencies, thereby reducing downtime and improving reliability.
[014] Yet another object of the present invention is to optimize not just individual
components but the entire compressor system. It ensures mass continuity, pressure
balance, and effective temperature management across all compression stages, resulting
in a more reliable and robust system that addresses the fundamental challenges of multi35 stage screw compressors.
D. SUMMARY OF THE INVENTION
[015] The various embodiments of the present invention provide an advanced modelling
framework for optimizing multi-stage screw compressors using Machine Learning (ML)
40 and Bayesian Optimization (BO). The invention integrates thermodynamic modelling,
4
5 predictive analytics, and optimization techniques to enhance the efficiency, reliability, and
performance of screw compressors across various industrial applications in real time. The
system includes two main tools: a Chamber Model, which follows the rules of physics to
predict how the compressor works, and a Machine Learning-based Gaussian Process
Regression (GPR) solver, which studies past data to make future predictions. By adjusting
10 things like pressure between stages, rotor shape, and oil injection, the system ensures the
compressor runs at its best while using less energy.
[016] According to one embodiment of the present invention, the multi-stage screw
compressor system consists of a Low-Pressure (LP) stage, an Intermediate Pipe, and a
High-Pressure (HP) stage. The compression process begins with the suction of air through
15 a filtration system, ensuring clean intake air. The LP stage performs the initial
compression, after which the air passes through the Intermediate Pipe, where it
undergoes cooling and additional oil injection before entering the HP stage for final
compression. The oil injection system in each stage provides lubrication, cooling, and
reduced mechanical friction, contributing to the overall efficiency of the system. The
20 compressed air then passes through an air-oil separator tank, where the injected oil is
removed and recirculated back into the system. The entire compression process is
powered by a diesel engine, ensuring continuous operation under varying load conditions.
[017] According to another embodiment of the invention, the optimization of the
compressor system is achieved using a combination of solvers. The Chamber Model is
25 responsible for simulating thermodynamic behaviour within the compressor stages,
accounting for variations in pressure, temperature, and fluid properties. The Gaussian
Process Regression (GPR) solver employs machine learning algorithms to analyse
historical performance data and predict optimal operating conditions in real-time. This
embodiment enables automatic adjustments to interstage pressure, rotor clearances, and
30 oil injection rates, ensuring maximum efficiency and minimal energy losses during
operation.
[018] According to yet another embodiment of the present invention, the system
incorporates a Graphical User Interface (GUI) that allows users to input parameters,
visualize real-time performance data, and select optimization solvers. The GUI provides
35 an interactive platform for compressor designers, engineers, and operators to fine-tune
performance settings and conduct multi-objective optimization. The framework supports
both single-objective and multi-objective optimization, allowing users to prioritize
different parameters such as volumetric efficiency, adiabatic efficiency, or specific power
consumption based on application-specific requirements. The system allows both single40 objective and multi-objective optimization, meaning users can focus on what matters
most—whether it’s energy efficiency, power savings, or better cooling.
[019] According to yet another embodiment of the present invention, the modelling
framework enables precise interstage pressure control, making it particularly useful for
5
5 high-pressure applications that require optimized energy efficiency and performance
stability. By predicting compressor performance trends and potential degradation, the
framework also enables predictive maintenance, reducing unplanned downtime and
improving the overall reliability of industrial compressor systems.
[020] According to yet another embodiment of the present invention, the modeling
10 framework is scalable and adaptable to different compressor configurations, including
single-stage, two-stage, and custom multi-stage systems.
[020] These and other aspects of the embodiments herein will be better appreciated and
understood when considered in conjunction with the following description and
accompanying drawings. It should be understood, however, that the following
15 descriptions, while indicating preferred embodiments and specific details, are given by
way of illustration and not limitation. Many changes and modifications may be made
within the scope of the embodiments herein without departing from the spirit thereof,
and the embodiments herein include all such modifications.
20 E. BRIEF DESCRIPTION OF DRAWINGS
[021] Fig. 1 illustrates a chart of of a two-stage compression screw compressor with
overlapping compression phases at intermediate pressure 𝑃6. Fig. 1 demonstrates the
relationship between pressure (P) and volume (V) across both stages, highlighting the
compression process from the initial suction pressure 𝑃+-. to the final discharge pressure
25 𝑃2.
[022] Fig. 2 illustrates the schematic representation of a two-stage oil-injected air screw
compressor.
[023] Fig. 3 illustrates the Two-stage oil-flooded air screw compressor setup used for
testing and validating the modelling framework, thereby illustrating the key components
30 and interconnections such as the low-pressure (LP) and high-pressure (HP) stages,
intermediate pipe, and associated systems.
[024] Fig. 4 illustrates a test setup for the two-stage oil-flooded air screw compressor,
featuring key components such as the suction filter, air-oil separator tank, two-stage air
end, and diesel engine for performance evaluation and validation of the modelling
35 framework.
[025] Fig. 5 illustrates a comparison of experimental data and Chamber Model predictions
for two stage screw compressors, illustrating the validation of the developed modelling
framework.
6
5 [026] Fig. 6 illustrates a comparison of Chamber Model Predictions with Gaussian Process
Regression (GPR) Predictions for total power consumption PtotalP_{total}Ptotal over
varying discharge pressures Pdis, including 95% confidence interval.
[027] Fig. 7 illustrates graphical user interface of the modelling framework.
[028] Fig. 8 illustrates a flowchart depicting the model architecture for multi-stage screw
10 compressor design, showing the progression of inputs, mapping, optimization, and
physics solver stages.
F. DETAILED DESCRIPTION OF THE INVENTION
[029] In the following detailed description, a reference is made to the accompanying
15 drawings that form a part hereof, and in which the specific embodiments that may be
practiced is shown by way of illustration. The embodiments are described in sufficient
detail to enable those skilled in the art to practice the embodiments and it is to be
understood that the logical, mechanical and other changes may be made without
departing from the scope of the embodiments. The following detailed description is
20 therefore not to be taken in a limiting sense.
[030] Various embodiments of the present invention provide for an advanced modelling
framework for multi-stage screw compressors.
[031] The following invention provides a process for optimizing the multi-stage screw
compressors and adjusts in real time the interstage pressure, rotor spacing, and oil flow
25 to get the best performance with the least energy waste.
[032] Multi-stage screw compressors operate by compressing gas through multiple
stages, with each stage increasing pressure incrementally to reach the desired output. The
optimization framework in this invention enhances this process by focusing on interstage
pressure adjustments, rotor profile improvements, and precise fluid injection strategies.
30 By optimizing these factors, the framework ensures that each compression stage operates
at its highest possible efficiency while reducing energy losses and minimizing mechanical
stress.
[033] One of the key aspects of multi-stage compression optimization is balancing
interstage pressures to reduce unnecessary energy expenditure. Conventional methods
35 rely on predetermined ratios, but the present invention uses real-time calculations and
optimization techniques to dynamically adjust pressure levels between stages, improving
overall compressor efficiency. Additionally, the system for the process refines rotor
geometry to reduce leakage and improve sealing efficiency, which is critical for
maintaining optimal pressure and airflow.
7
5 [034] Fluid injection plays a crucial role in managing heat and lubrication within multistage compressors. This framework optimizes fluid injection rates at strategic points
within the system to ensure effective cooling and lubrication, preventing excessive wear
on mechanical components. By incorporating these advanced techniques, the
optimization framework enhances compressor performance while reducing power
10 consumption and operational costs.
[035] This system and the process provides two types of solvers for optimization: a
physics-driven Chamber Model and a machine learning-based Gaussian Process
Regression (GPR) solver.
[036] The Chamber Model relies on thermodynamic equations to simulate compressor
15 behavior, while the GPR solver leverages data-driven learning to make accurate
predictions. Users can choose between these solvers or use them together, depending on
their needs. The integration of machine learning introduces a predictive capability that
enhances the system’s adaptability and efficiency.
[037] The system and process require key operational inputs such as suction pressure,
20 discharge pressure, suction temperature, rotor speed, and the type of working fluid.
These parameters define the operating conditions of the compressor and influence its
efficiency and performance. By allowing users to fine-tune these variables, the system can
adapt to different industrial applications and environmental conditions.
[037] To evaluate the efficiency of the compressor, the framework measures critical
25 performance indicators such as volumetric efficiency, adiabatic efficiency, and specific
power consumption. These metrics help users understand how effectively the compressor
converts input energy into useful work while minimizing energy waste.
[038] The system optimizes rotor profile clearances, including interlobe, axial, and radial
clearances. These clearances are crucial because they influence leakage rates and overall
30 compressor efficiency. Additionally, parameters like the built-in volume ratio and wrap
angle are optimized to ensure minimal energy losses and maximum performance. By
refining these geometrical aspects, the system enhances the compressor's reliability and
lifespan.
[039] The fluid injection system in a multi-stage screw compressor plays a crucial role in
35 optimizing performance, cooling, lubrication, and pressure regulation. Unlike
conventional fluid injection systems that operate at fixed rates, this invention introduces
a dynamic and intelligent fluid injection strategy that adjusts in real time based on
operating conditions.
[040] Screw compressors generate significant heat due to the compression process and
40 friction between moving components. Uncontrolled heat buildup can reduce efficiency,
8
5 cause excessive wear on mechanical parts, and even lead to system failure. Fluid injection
helps manage these issues by providing:
Cooling – Injected fluid absorbs excess heat generated during compression, preventing
overheating.
Lubrication – The fluid creates a thin layer between moving parts, reducing friction and
10 mechanical wear.
Pressure Regulation – Properly managed fluid injection helps maintain consistent
pressure levels throughout different stages of compression, improving efficiency.
[041] The system dynamically regulates the amount of injected fluid based on factors like
temperature, pressure, and rotor speed. Instead of a fixed-rate injection, sensors
15 continuously monitor compressor conditions, and the framework optimizes fluid injection
accordingly.
[042] Fluid is introduced at multiple critical points, including the low-pressure stage,
interstage region, and high-pressure stage. This ensures that each compression stage
receives adequate cooling and lubrication, enhancing overall performance.
20 [043] Injection parameters such as fluid type, temperature, pressure, and rate are
optimized for maximum efficiency. The framework uses Bayesian optimization and
machine learning to determine the most effective injection settings for different
operating conditions.
[044] By reducing excessive heat and mechanical stress, the system extends the
25 compressor’s lifespan. It also lowers energy consumption by minimizing heat-related
losses, making the compressor more energy-efficient.
[045] The system includes an interactive and user-friendly GUI that simplifies parameter
input, solver selection, and optimization execution. This feature makes it accessible to
both experienced engineers and new users by allowing them to interact with the system,
30 run simulations, and visualize optimization results in real-time.
[040] This system supports both single and multi-objective optimization. Users can finetune performance factors such as energy efficiency, cooling effectiveness, and overall
power consumption simultaneously. The system leverages advanced algorithms like
Bayesian optimization and genetic algorithms to balance multiple factors and achieve the
35 best possible compressor performance.
[041] In the context of the multi-stage screw compressor system, the function or purpose
of each part is as follows:
1. Low Pressure (LP) Stage (1):
9
5 The LP stage is the first compression stage where the working fluid (e.g., air) enters the
system. At this stage, the fluid is compressed to an intermediate pressure before being
passed to the next stage. This stage is responsible for initial compression from suction
pressure to the intermediate level.
2. Intermediate Pipe (2):
10 The intermediate pipe connects the low-pressure stage to the high-pressure stage. It
allows the compressed fluid from the LP stage to flow into the high-pressure stage for
further compression. Additionally, it serves as an intercooler section where oil is injected
to control temperature and ensure efficient heat transfer between the stages.
3. High Pressure (HP) Stage (3):
15 The HP stage further compresses the fluid that has been pre-compressed in the LP stage.
It increases the pressure of the working fluid to the final discharge pressure. This stage
completes the compression process, delivering the required high-pressure output for the
specific application.
4. Suction Filter:
20 The suction filter is installed at the system's intake to ensure that the incoming air or gas
is free from contaminants such as dust, debris, or other impurities. This ensures that clean
fluid enters the compressor stages, preventing damage to internal components and
maintaining system efficiency.
5. Air-Oil Separator Tank:
25 This component separates the oil from the compressed air after the compression process.
The oil, which is injected into the compressor for lubrication and cooling purposes, is
removed from the compressed air before it is delivered to the output, ensuring clean and
dry air is supplied to the end application.
6. Diesel Engine:
30 The diesel engine provides the necessary mechanical power to drive the screw
compressor. It converts fuel energy into mechanical energy, which is then transmitted to
the compressor rotor system to initiate and maintain the compression process. The diesel
engine is particularly useful for mobile or off-grid applications.
7. 2-Stage Airend:
35 The 2-stage airend consists of the rotor and housing assembly for both the low-pressure
and high-pressure stages. It houses the screw rotors, which are the core mechanical
elements responsible for compressing the air as it moves through the system. The 2-stage
10
5 design allows for improved efficiency by distributing the compression across two stages
instead of one.
Together, these components are interconnected to form a comprehensive system capable
of efficiently compressing air or gas across multiple stages, ensuring high-pressure output
while maintaining temperature control and efficiency.
10 [042] The operation of the multi-stage screw compressor using the advanced modeling
framework is a systematic process involving the interaction between various components
of the compressor system and the integration of optimization techniques to achieve
efficient performance. The steps below describe the operational flow of the invention:
1. Suction Process: The working fluid (e.g., air) enters the system through the suction filter,
15 which ensures the air is clean and free from particulates. This air enters the Low Pressure
(LP) Stage (1), where the first stage of compression occurs.
2. Compression in the Low-Pressure Stage: In the LP stage, the air is compressed by the
rotation of the screw rotors housed in the 2-Stage Airend. The rotors create a volume
reduction as the air passes through, raising its pressure to an intermediate level. During
20 this stage, oil is injected into the compression chamber to reduce heat generated by
compression, lubricate the rotors, and prevent excessive wear.
3. Intercooling and Intermediate Pipe Transfer: The compressed air from the LP stage
flows into the Intermediate Pipe (2), where it undergoes further cooling to reduce the
temperature before entering the high-pressure stage. Additional oil is injected into this
25 pipe to assist with cooling and maintaining optimal conditions for the second stage of
compression.
4. High-Pressure Compression: The cooled and compressed air from the intermediate pipe
enters the High Pressure (HP) Stage (3). Here, the air undergoes further compression,
raising its pressure to the final discharge level. Similar to the LP stage, oil injection is
30 applied during this stage to maintain temperature control, ensure smooth rotor
operation, and minimize friction.
5. Oil-Air Separation: After the air has been compressed to the desired high pressure, it
passes through the Air-Oil Separator Tank. The injected oil is separated from the
compressed air, ensuring that clean air exits the system for the intended application, while
35 the oil is recirculated back into the system for reuse.
6. Powering the Compressor: The entire compression process is powered by a Diesel
Engine, which drives the screw rotors in both the LP and HP stages. The engine provides
the necessary mechanical energy to rotate the rotors, ensuring continuous air
compression.
11
5 7. Optimization and Control: The operational efficiency of the compressor is enhanced by
the modeling framework, which integrates two key solvers: the Chamber Model and the
Gaussian Process Regression (GPR). These solvers allow for real-time optimization of key
parameters such as the interstage pressure, rotor clearances, and fluid injection rates. The
modeling framework ensures that the compression process is optimized for energy
10 efficiency, minimizing power consumption while achieving the desired output pressure.
The Chamber Model simulates the thermodynamic behaviour of the compression process,
calculating the pressure and temperature changes as the air passes through the
compressor stages.
Gaussian Process Regression (GPR) Solver: A machine learning-based solver that uses
15 statistical data to predict the performance and optimize parameters for improved
efficiency.
8. Discharge: After compression, the high-pressure air exits the HP stage, ready to be used
in various applications, such as water well operations, industrial processes, or other highpressure air requirements.
20 In summary, the operation of the multi-stage screw compressor involves sequential
compression through the LP and HP stages, with optimization of critical parameters using
advanced modeling techniques to ensure high efficiency, precise pressure control, and
reduced energy consumption.
[043] The operation of the multi-stage screw compressor using the advanced modeling
25 framework requires several key elements, each playing a vital role in the successful
functioning and optimization of the compressor. The basic elements required for the
operation include:
1. Screw Compressor (Two-Stage) Components:
Low Pressure (LP) Stage (1): The first compression stage, where the air is compressed to
30 an intermediate pressure level.
High Pressure (HP) Stage (3): The second compression stage, where the air is compressed
to its final discharge pressure.
Intermediate Pipe (2): Connects the LP and HP stages, acting as a conduit for the
compressed air while allowing for intermediate cooling and oil injection.
35 Rotors (Male and Female): The rotating screws inside both the LP and HP stages, which
create the compression by reducing the air volume as it moves through the rotors.
Oil Injection System: Oil is injected into both stages and the intermediate pipe to cool the
air, lubricate the rotors, and reduce friction.
12
5 2. Diesel Engine:
The engine provides the mechanical power required to rotate the screw rotors in both the
LP and HP stages. This element is crucial for driving the compression process.
3. Air-Oil Separator Tank:
This element separates the oil from the compressed air after the compression process,
10 ensuring that the air delivered to the system is free of oil, and allowing the oil to be
recirculated back into the compressor.
4. Suction Filter:
The filter ensures that only clean air enters the system, free of dust and other particulates,
which could damage the internal components of the compressor.
15 5. Advanced Modeling Framework:
Chamber Model Solver: A physics-based solver that simulates the thermodynamic
behavior of the air inside the compression chambers.
Gaussian Process Regression (GPR) Solver: A machine learning-based solver that uses
statistical data to predict the performance and optimize parameters for improved
20 efficiency.
6. Graphical User Interface (GUI):
The GUI allows users to interact with the modeling framework, input various parameters
such as pressure, temperature, and rotor speed, and choose optimization functions to
improve the compressor's performance.
25 7. Oil Injection System:
The oil injection system ensures proper cooling and lubrication of the compressor during
the compression stages. It helps control the temperature of the air and prevents excessive
wear on the rotors.
8. Interstage Pressure Control:
30 This element ensures that the pressure between the LP and HP stages is optimized for
maximum efficiency, allowing for balanced compression and reduced energy
consumption.
9. Control and Measurement Sensors:
Pressure, temperature, and mass flow sensors are critical for monitoring the compressor's
35 performance and feeding real-time data into the modeling framework for optimization.
13
5 10. Cooling System:
A cooling system is required to ensure that the compressor operates within safe
temperature limits, preventing overheating, and maintaining the desired thermal
conditions.
These elements work together to ensure the proper functioning, monitoring, and
10 optimization of the multi-stage screw compressor, making it an efficient and reliable
system for industrial and high-pressure applications.
[044] Example 1: Two-Stage Oil-Flooded Air Screw Compressor
1. Setup Overview: The system comprises a two-stage screw compressor, connected to a
diesel engine for power, with an air-oil separator tank and a suction filter integrated to
15 ensure efficient compression of air. Oil injection is provided at three critical points—
within the first stage (LP), the intermediate pipe, and the second stage (HP)—to ensure
proper lubrication and cooling throughout the compression process.
2. Operating Conditions:
Suction Pressure (P₁): 0.95 bar (atmospheric pressure)
20 Suction Temperature (T₁): 300 K
Discharge Pressure (P₄): 25.07 bar (final stage discharge pressure)
Intermediate Pressure (P₃): Optimized to 4.68 bar using the modelling framework for
maximum efficiency.
Male Rotor Speed (LP & HP): 3026 RPM (LP) & 3713 RPM (HP)
25 Working Fluid: Air
3. Optimization Parameters: Using the modelling framework and its integrated solvers,
the following parameters were optimized:
Intermediate Pressure: The chamber model solver calculated the optimum intermediate
pressure between the LP and HP stages to be around 4.68 bar, balancing efficiency and
30 power consumption.
Fluid Injection Parameters: The oil injection rates and temperatures were adjusted to
ensure effective cooling and minimize friction losses.
Rotor Profile Clearances: The interlobe and radial clearances were minimized to reduce
leakage and enhance volumetric efficiency.
35 4. Results:
14
5 The total power consumption of the compressor was reduced from 315 kW (without
optimization) to 307 kW (with optimization), representing a significant improvement in
energy efficiency.
The volumetric efficiency was improved by 3%, resulting in more effective air
compression per unit of energy input.
10 Discharge Temperature: The discharge temperature at the final stage was reduced,
preventing overheating and improving the overall reliability of the system.
5. The graph (shown in Figure~ 5 & 6) compares the experimental results and chamber
model predictions, showing close agreement between the two, with deviations as low as
2.44% to 3.02%, validating the model's accuracy.
15 6. Application to Product: This optimized two-stage screw compressor, configured using
the developed modeling framework, was tested in a water well application, where it
demonstrated improved performance and efficiency. By adjusting parameters such as
interstage pressure, rotor profile clearances, and oil injection rates, the compressor's
output was maximized while keeping energy consumption low.
20 Example 2: Multi-Stage Airend Testing Setup
A KPCL two-stage airend connected to a diesel engine was tested under various operating
conditions to validate the modeling framework’s predictions for larger-scale industrial
applications. The testing setup included:
Suction Filter: Ensuring only clean air enters the system.
25 Air-Oil Separator Tank: Efficiently separating oil from the compressed air.
Intermediate Pipe: Acting as a conduit between stages, with optimized oil injection.
Using the framework’s Bayesian optimization solver, the system achieved a balance
between power consumption, flow rate, and pressure ratios. In this example, the
framework allowed the user to choose specific performance metrics, such as adiabatic
30 efficiency and total power consumption, for multi-objective optimization.
In both working examples, the advanced modelling framework improved efficiency,
reduced power consumption, and provided optimized performance metrics for the
respective applications. The system's flexibility in selecting optimization paths and
solvers made it adaptable to both small-scale and large-scale operations.
35 [045] Key innovations in the present invention include:
1. Hybrid Modelling Framework
15
5 Traditional screw compressor models rely on either purely physics-based calculations or
empirical data-driven approaches. This invention introduces a hybrid framework that
integrates a Chamber Model (for thermodynamic calculations) with a Machine Learningbased Gaussian Process Regression (GPR) solver. This allows users to either select one
approach or combine both, offering unprecedented flexibility in compressor modelling.
10 The unique ability to switch between a physics-driven solver and a machine learningbased solver, or blend the two for improved accuracy, marks a significant advancement
in screw compressor optimization.
2. Bayesian Optimization for Multi-Stage Compressors
Conventional optimization methods rely on trial-and-error or simplistic optimization
15 techniques, which can be time-consuming and inefficient. This invention integrates
Bayesian Optimization, enabling efficient exploration and fine-tuning of critical
parameters such as interstage pressure, rotor geometry, and fluid injection rates. This
leads to faster, more precise, and highly efficient optimization.
The multi-objective capability of Bayesian Optimization ensures that multiple factors—
20 such as efficiency, cooling, and power consumption—are simultaneously optimized,
making the process both faster and more effective.
3. Interstage Pressure Optimization
Standard screw compressor designs use the geometric mean method to estimate
interstage pressure. However, this method does not account for real-world inefficiencies
25 like friction and heat exchange losses. This invention dynamically adjusts interstage
pressure in real-time, resulting in more accurate modelling and enhanced compressor
performance.
The adaptive real-time optimization of interstage pressure, considering real-world losses,
allows the system to be highly responsive and efficient, particularly in high-pressure
30 industrial applications.
4. Fluid Injection and Cooling Optimization
Most screw compressors apply fixed-rate fluid injection, which often results in inefficient
cooling and lubrication. This invention dynamically adjusts fluid injection parameters—
including pressure, temperature, and flow rate—according to the specific needs of each
35 compression stage. This targeted approach enhances cooling, lubrication, and overall
efficiency.
The real-time optimization of fluid injection ensures better thermal management,
reduces mechanical wear, and extends the compressor’s operational lifespan while
enhancing performance.
16
5 5. Multi-Objective Optimization Capabilities
Traditional optimization methods focus on improving only one performance factor at a
time, such as energy efficiency or volumetric efficiency. This invention supports multiobjective optimization, allowing users to simultaneously balance multiple parameters
such as power consumption, volumetric efficiency, and cooling performance.
10 The framework’s ability to perform both single and multi-objective optimization,
leveraging advanced algorithms like Bayesian Optimization and Genetic Algorithms,
provides users with highly customizable compressor designs tailored to specific industrial
needs.
6. User-Friendly Graphical User Interface (GUI)
15 Many existing compressor modeling tools have complex user interfaces with limited
parameter control. This invention includes an intuitive GUI, simplifying input
management, solver selection, and real-time optimization execution. The user-friendly
interface ensures that both experienced engineers and new users can easily interact with
and leverage the system’s full potential.
20 The interactive GUI seamlessly integrates advanced optimization techniques, making
sophisticated compressor modeling accessible to a wider range of users without requiring
deep technical expertise.
7. Experimental Validation with Real-World Data
Many theoretical models for screw compressors lack validation with actual experimental
25 data, making their real-world applicability uncertain. This invention has undergone
extensive experimental validation, demonstrating a low error margin (2.44% to 3.02%)
between predicted and real-world performance results.
The high accuracy of the model’s predictions, validated against real-world data, ensures
reliability and trustworthiness, setting it apart from conventional theoretical models that
30 may lack empirical testing. , Claims:We Claim:
1. A system and process for optimization of multi-stage screw compressors comprising:

A modelling framework utilizing a physics-based Chamber Model, which applies thermodynamic principles to simulate screw compressor performance based on heat transfer, gas dynamics, and pressure-volume relationships;

A machine learning-based Gaussian Process Regression (GPR) solver, trained on prior experimental or simulation data, to provide compressor performance predictions without the need for direct physical modelling;

A parameter optimization system for refining operational variables such as interstage pressure, rotor geometry, and fluid injection rates based on historical data and empirical models; and

A Graphical User Interface (GUI) that enables users to input compressor parameters, execute simulations, and visualize performance outputs

characterized by

a hybrid modelling framework, enabling the user to switch between the Chamber Model and GPR solver or combine them dynamically, wherein the hybrid approach leverages the Chamber Model for physics-based accuracy and the GPR solver for rapid predictions, with cross-validation ensuring enhanced reliability;

a Bayesian Optimization module, integrated into the parameter optimization system, which actively fine-tunes interstage pressure, rotor geometry, and fluid injection rates based on a probabilistic learning process, enabling more precise and computationally efficient optimization compared to conventional methods;

a real-time interstage pressure control mechanism, dynamically adjusting pressure based on actual thermodynamic losses such as heat dissipation and friction, moving beyond traditional fixed-ratio pressure estimations;

a fluid injection optimization system, wherein injection pressure, temperature, and flow rate are actively adjusted in response to compressor load variations, ensuring efficient cooling, lubrication, and minimized mechanical wear;

a multi-objective optimization module, allowing simultaneous optimization of multiple performance metrics—such as volumetric efficiency, power consumption, and cooling effectiveness—by applying Bayesian Optimization and Genetic Algorithms for intelligent trade-off analysis; and

a Graphical User Interface (GUI) enhancement, integrating real-time comparative analysis, cross-validation tools for hybrid solver selection, and interactive visual feedback to facilitate rapid decision-making in compressor optimization.

2. The system of claim 1, wherein the Chamber Model simulates the thermodynamic behaviour of the screw compressor by applying equations governing heat transfer, gas dynamics, and pressure-volume relationships.

3. The system of claim 2, wherein the Chamber Model is adapted to incorporate rotor profile clearances, including interlobe, axial, and radial clearances, to enhance predictive accuracy.

4. The system of claim 1, wherein the Gaussian Process Regression (GPR) solver predicts compressor performance based on historical simulation data, enabling faster optimization without requiring extensive physical modelling.

5. The system of claim 4, wherein the Gaussian Process Regression (GPR) solver refines its predictive accuracy over time by learning from additional simulation and real-world experimental data.

6. The system of claim 1, wherein the hybrid modelling framework enables the simultaneous execution of the Chamber Model and GPR solver, with the ability to cross-validate predictions for improved accuracy.

7. The system of claim 6, wherein the hybrid framework cross-validates physics-based and machine learning predictions, selecting the more accurate output based on an error minimization function.

8. The system of claim 1, wherein the Bayesian Optimization module repeatedly refines key parameters based on past optimization results to minimize energy losses and enhance compressor efficiency.

9. The system of claim 8, wherein the Bayesian Optimization module is configured to perform gradient-free optimization, making it adaptable to complex, nonlinear compressor performance landscapes.

10. The system of claim 1, wherein the interstage pressure control mechanism dynamically recalculates pressure based on real-time operating conditions, including heat losses, friction, and compression ratio variations.

11. The system of claim 10, wherein the real-time interstage pressure control mechanism is coupled with a sensor-based feedback system that continuously monitors actual pressure levels and automatically adjusts setpoints to ensure optimal performance.

12. The system of claim 1, wherein the fluid injection optimization system adjusts injection parameters in response to compressor load, rotational speed, and thermal conditions, preventing overheating and excessive wear.

13. The system of claim 12, wherein the fluid injection optimization system accounts for changes in gas composition and moisture levels, preventing excessive condensation or dry operation.

14. The system of claim 1, wherein the multi-objective optimization module allows users to specify priority objectives such as maximizing cooling efficiency while minimizing power consumption, providing a customizable optimization approach.

15. The system of claim 14, wherein the multi-objective optimization module includes a genetic algorithm-based solver, ensuring optimal trade-offs between power efficiency, cooling capacity, and mechanical reliability.

16. The system of claim 1, wherein the Graphical User Interface (GUI) includes:

A parameter input module allowing users to configure compressor operating conditions;

A solver selection module enabling switching between the Chamber Model, GPR solver, or hybrid mode;

A real-time visualization module displaying optimization progress, performance predictions, and suggested operational improvements; and

A user control panel for executing optimizations and adjusting system constraints.

17. The system of claim 16, wherein the Graphical User Interface (GUI) further includes a comparative analysis tool, allowing users to visualize and compare different optimization scenarios before finalizing compressor settings.

18. A method for optimizing a multi-stage screw compressor, comprising:
receiving input parameters, including suction pressure, discharge pressure, temperature, rotor speed, and fluid type;

executing a hybrid modelling approach, using a Chamber Model and/or Gaussian Process Regression (GPR) solver to simulate compressor performance;

applying Bayesian Optimization to refine interstage pressure, rotor geometry, and fluid injection rates;

dynamically adjusting interstage pressure based on real-time heat loss and frictional considerations;

optimizing fluid injection parameters, including pressure, temperature, and injection rate, in response to compressor operating conditions;

performing multi-objective optimization to balance efficiency, cooling, and power consumption; and

presenting optimization results to the user through an interactive Graphical User Interface (GUI).

19. The method of claim 10, wherein the Bayesian Optimization step further comprises an adaptive learning function that prioritizes parameter adjustments based on historical performance data.

20. The system of claim 3, wherein the Gaussian Process Regression (GPR) solver refines its predictive accuracy over time by learning from additional simulation and real-world experimental data.

21. The method of claim 10, wherein the fluid injection optimization step dynamically adjusts injection timing based on compressor load variations, ensuring minimal energy losses.