Environment & Safety Gas Processing/LNG Maintenance & Reliability Petrochemicals Process Control Process Optimization Project Management Refining

November 2023

Water Management

Energy savings through the renewal of cooling tower fan blades

The cooling water system in a refinery consists of four loops that can be operated independently.

Tüpraş: Oğuş, E.

The cooling water system (CWS) in a refinery consists of four loops that can be operated independently. Each loop is linked to a cooling tower that is similar conceptually but has different capacities (from two to five cells). Each of these cells is equipped with a fan that can be manually switched on or off to optimize cooling performance. Every cooling water (CW) loop has dedicated CW pumps (from two to four pumps) to maintain a proper flowrate and manipulate the system pressure. No variable-speed drive is applied on the CW pumps, so the only “control” for flowrate and pressure is to switch the pumps on or off.

There are different interconnections between the four CW loops that allow the CWS to operate flexibly, and which ensure a high reliability of CW supply. During periods with reduced CW requirements (winter season, plant shutdowns), it may be possible to switch off one CW cycle completely. FIG. 1 shows an overview of the Izmit refinery’s CWS. The different loops and supply alternatives to units can be identified by the differently colored piping system.

FIG. 1. Cooling tower configuration.

The main technical data of the cycles are listed in TABLE 1. The data show that the CW pumps and fans are substantial energy consumers, so it is important to operate them in the most efficient way.

It was decided to replace cooling tower’s fan blades at with hollow e-glass epoxy fiber-reinforced plastics (FRP) composite fan blades as designed on the principles developed by a well-known institution and supplied by a reputed Indian solution providera.

The function of a cooling system is to remove heat from a process or equipment. Heat removed from one medium is transferred to another medium or process fluid. Most often, the cooling medium is water in places where water availability is not a constraint.The transfer of heat from process fluid or equipment results in a rise in temperature in the CW. In an open recirculating system, cooling is achieved through the evaporation of a fraction of the water. An open recirculated system uses the same water repeatedly to cool process equipment (FIG. 2). These systems are designed to cool water to a certain temperature under a given set of conditions. 

FIG. 2. Cooling tower process.

Counterflow induced draft towers (FIG. 3) are a type of tower in a refinery where the working principle of the fan is to create vertical air movement up the tower and across the packing media in opposition to the water flow. Therefore, the coldest water contacts the driest air first. Considering construction and the selection of materials, mechanical draft towers may be examined as components under the headings of water tank, water distribution, casing, fan and packing.

FIG. 3. Induced draft counterflow tower.

Traditional design guidelines dictate that fans should be designed to operate successfully in cooling tower service for at least 5 yr. The fans should have a minimum of four to six blades of aluminum alloy, stainless steel, monel or an equal blend. Blades should be adjustable for degree of pitch and individually fastened to a hub of welded steel or cast iron (stainless steel optional at extra cost).

Efficient modern fan design focuses on the following factors:

  • Selection of the optimum aerofoil cross-sectional profile
  • Optimization of the fan blades to the required duty
  • Selection of raw materials
  • Combination of raw materials
  • Manufacturing process.

Aerospace engineering scientists have been involved in the aerodynamic design development of FRP fans and a strong body of knowledge has been developed by the aerospace industry. Technical consultations with scientists were involved with primary aeronautical designs. While fan performance depends on material of construction, fan diameter and plenum optimization, the number of blades and their efficiency can be demonstrated by lower power consumption. Fan performance is dependent on airfoil section, blade profile at different radii and staggering, chord length at different radii, thickness, camber, blade twist, angle of attack and the  smooth surface finish of the blades, among many more parameters. The design of the hub can help minimize air cycling loss, and selecting the number of blades after consideration of the above parameters determines the setting of the required air volume and pressure.

While fan revolutions per minute (rpm) are linked to tip speed, noise and vibration, most of a fan’s performance relies on fan design and deterioration. Pressure distribution on the airfoil is one of the main components of performance. Faster airflow causes a decrease in air pressure. Air flow increases over the curved upper surface if the wing speed increases. Increased speed reduces the pressure above the wing and produces the upward lifting force (Eq. 1):

ղ = Q × TP / KW                 (1)


Q = Airflow, m3/sec

TP = Total pressure (Pascal)

ղ = Fan efficiency

Fan efficiency formulae

Many criteria factor into material selection: corrosion resistance to acidic or alkaline water, higher efficiency due to improvement,  aerodynamics characteristics, structural integrity, high strength-to-weight ratio, and general adaptability to complex geometries are key parameters. The main types of materials are epoxy resins and polyesters. Epoxy resins are prefered as they provide higher mechanical strength, improved chemical resistance, negligible water absorbption and superior heat resistance. Polyesters are a weak resin structure and have high water absorption. E-glass composite is the ideal glass fabric of an axial flow fan as it provides higher tensile, shear and cross-breaking strength after being bonded with the epoxy resins.

The material used in this project offers more ultimate tensile and compressive strength than even steel. The blade material is defined as e-glass fiber-reinforced polymer with a high-grade epoxy resin. A higher lift-to-drag ratio and aerodynamic design are districtive chacteristics. The chosen blades are uniform in weight, shape and balance, provide for manual adjustment of the pitch angle, and are easily removable. They provide uniform air velocity from hub to tip with low noise and vibration. Each fan blade is statically balanced and the hub dynamically balanced at the factory for quieter and more efficient operation.

Process staff and related professionals carried out the testing in accordance with international standards and code for a field performance test for cooling tower fans. The international standard measurement method specifies that measurement points should be measured on the hood with the anemometer used to measure the air flow.

This kind of test requires mild wind conditions, as the standard dictates that wind velocity should be less than the average fan outlet velocity.

Air flowrates before and after the execution of the replacement were kept constant by adjusting the blade angles. Air flowrates were measured while recording the energy consumptions. As discussed above, an approved anemometer was used. Linear velocity was measured with this instrument and the air flow was calculated by multiplying that velocity by the area. All air velocity measurements were taken from the lowest point of the fan hood with the help of the pier.

Data was collected from 20 points for some 20 min. After the fans were replaced with new, high-efficient FRP fans, all measurements were repeated. Scaffolding was required around the hood to make flow measurements.

Good test results cannot be obtained in greater wind velocities. It is also not recommended to perform the test if wind conditions exceed 9 mph–10 mph. İf the wind shifts direction during the time of flow traverse, the test should be paused. It is also known that winds higher than 14 mph–15 mph above average cause inconsistent test measurement values. The delivered air volume can be calculated from the anemometer velocity reading and is multiplied by the area of the fan stack discharge. FIG. 4 summarizes the manner of anemometer test positions. An acceptable reading number is around 20.

FIG. 4. The manner of anemometer test positions.

An electronic anemometer can directly read velocity with high accuracy and moisture resistance. It can display velocity and—for each measurement quarter—calculate the average of measurements. Due to the considerable amount of time required, “yaw” (horizontal and vertical velocities) angle measurements may not be performed as in this project.

The fan supplier company guaranteed a minimum 18% (± 3%) increase in the air flow performance at the same or lower power consumption of the existing operating cooling tower fan after the replacement with a high-efficiency aerodynamic design e-glass epoxy FRP fan. The company also guaranteed at least a 25% (± 5%) reduction in the power consumption performance at the same operating airflow of the existing operating cooling tower fan assembly after the replacement with the high-efficiency aerodynamic-designed e-glass epoxy FRP fan assembly.

The current conventional fan blades were replaced with new fan blades in four cells in the 9E-501 tower, in five cells in the 9E-601 tower, in three cells in the 10E-103 tower and in two cells in the 17E-201 tower. In total, 14 sets of existing fan blades were replaced with a one-blade addition for each set. The project was implemented from March 2021–August 2021.

Sample details of calculations and measurements

TABLE 2 summarizes the energy consumption results of the Tower 1 fan blade change. TABLE 3 shows the energy measurement values before and after the fan blade replacement.


To calculate the fan area, the following values were taken for the diameter of the point where the speed measurement was made and the diameter of the hub. Smaller hub diameters create additional advantage for the air passage rate:

  • Diameter of the circle at the measuring point on the fan: 8 m
  • Old hub diameter: 1.4 m
  • New hub diameter: 1.2 m
  • Fan upper ring area: π*(82)/4 = 50.26 m2
  • Old hub area: π*(1.42)/4 = 1.54 m2
  • Fan old area: 50.26 m2 –54 m2 = 48.72 m2
  • New hub area: π*(1.22)/4 = 1.13 m2
  • Fan new area: 50.26 m2 –13 m2 = 49.13 m2
  • Power saving = (86-66.48)/86 x 100 = 22.6%
  • Flow before = 8.071 m/sec x72 m2 x 3,600 s/hr = 1,415.588 m3/hr
  • Flow after = 8.045 m/sec x13 m2 x 3,600 s/hr =1,422.903 m3/hr
  • Only 0.5% increase in airflow.

TABLE 4 shows the energy consumption data as a result of the Tower 2 fan blade change. TABLE 5 shows the energy consumption data as a result of the Tower 3 fan blade change. TABLE 6 shows the energy consumption data as a result of the Tower 4 fan blade change. TABLE 7 shows the air flowrates before and after replacement (m3/hr), and TABLE 8 shows the average fan outlet air velocity rates (m/sec). FIG. 5 shows the Cooling Tower 9E-501 hot water return from process (red) and cold water supply to process (blue) temperatures. No unexpected issues were shown for 2 yr.


FIG. 5. The Cooling Tower 9E-501 hot water return from process (red) and cold water supply to process (blue) temperatures.
FIG. 5. The Cooling Tower 9E-501 hot water return from process (red) and cold water supply to process (blue) temperatures.


After the project’s implementation, extensive field tests confirmed that the energy savings—for equal performance—averaged 25%–30%. No process issues have been observed after the installations, and the fans have been working as intended.

Energy savings as well as improved equipment safety and control can be expected by imlementing this technology in process plants. Distillation column overhead fin fan cooler condensers are an ideal place to implement this technology to improve heat duty, improve mass transfer efficiency, reduce flaring and significantly improve gross refinery margins. HP



a Encon Group

The Author

Related Articles

From the Archive



{{ error }}
{{ comment.name }} • {{ comment.dateCreated | date:'short' }}
{{ comment.text }}