Experimental Study of Flow-induced Pressure Fluctuations in Horizontal Straight Pipes with Single- and Two-phase Flow
Article information
Abstract
Pressure fluctuations caused by multiphase flow in piping systems can lead to structural vibration, noise generation, and potential damage to the piping infrastructure. This study examined flow-induced pressure fluctuations and their spatial distribution on the pipe walls for single-phase and two-phase flows in a horizontal straight pipe. The experiments were conducted using a flow test loop under various flow velocities and water fractions. A simplified one-dimensional numerical model was used to examine the influence of a reservoir tank on the pressure magnitude and distribution within the piping system. The results showed that pressure fluctuations in single-phase flow are governed primarily by the flow velocity, whereas those in two-phase flow are strongly influenced by water fraction and free-surface deformation. Comparisons between open-loop and closed-loop conditions showed that the reservoir tank induces low-frequency oscillations that disappear in the closed-loop configuration. These findings provide fundamental experimental data for understanding flow-induced pressure fluctuations and their implications for vibration and noise in piping systems used in marine and offshore engineering applications.
1. Introduction
Piping systems are used widely in various engineering systems, including ships, offshore structures, and industrial plants, where they often handle multiphase flows consisting of gas and liquid. Multiphase flow in piping systems can induce significant pressure fluctuations along the pipe walls, leading to fluid-induced vibration (FIV), which may cause structural deformation and fatigue failure (Belvins, 1977). These systems experience complex fluid dynamics influenced by factors such as geometry, components (e.g., elbows, pumps, valves, and orifices), and the varying amplitudes and frequencies of the flow (Nakamura et al., 2013). Such pressure fluctuations pose critical challenges for ships and offshore structures, where piping networks are used widely to transport fluids such as cooling water, fuel, lubricating oil, and ballast water. Flow-induced pressure fluctuations in these systems can generate vibrations and noise that may affect structural integrity and onboard acoustic environments.
Several studies have investigated the characteristics of multiphase flow in piping systems, focusing on the resulting vibrations and noise. Riverin and Pettigrew (2007) examined the vibration mechanisms caused by two-phase flow in straight and elbow pipes. They showed that the primary vibration frequency of the pipe was associated with the periodic characteristics of two-phase flow, underscoring the role of flow regimes in generating structural vibrations. Belfoid (2010) conducted experimental assessments of the forces and accelerations induced by two-phase flow in straight pipes, bends, and joints. They reported that flow-induced forces were significantly influenced by the flow regime, with two-phase flow generating significantly higher forces than single-phase flow. Su et al. (2022) extended these investigations, focusing on two-phase flow-induced vibrations in bend pipes. They showed that the internal momentum flux of the flow strongly affects lateral and axial vibrations, while vertical vibrations are governed by the flow regime.
In addition to experimental studies, numerical approaches using computational fluid dynamics (CFD) have also been used to analyze the flow-induced loads and noise in piping systems. Hambric et al. (2010) examined the structural and fluid-borne acoustic power generated by single-phase flow in a 90° elbow pipe. Using CFD coupled with structural-acoustic models, they proposed empirical formulae to predict the sound levels based on the turbulent flow characteristics. Similarly, Wang et al. (2018) used a direct numerical simulation (DNS) to study turbulent characteristics in 90° elbow pipes, identifying swirl-switching phenomena and evaluating the measurement techniques for analyzing complex flow behaviors.
Previous studies have focused primarily on evaluating the forces, vibrations, and noise caused by single- and two-phase flow in pipes using experimental methods and numerical simulations. Although these studies provide valuable insights, they have largely focused on specific geometries, such as bends and elbows, often neglecting straight pipes, which are fundamental elements of piping systems. Eguchi et al. (2011) used a CFD-based large-eddy simulation (LES) to investigate the pressure fluctuations caused by single-phase internal flow in elbow pipes. Their findings highlighted a strong correlation between the pressure fluctuations and forces acting on the pipe wall, but the study was limited to elbow geometries, leaving the behavior in straight pipes underexplored. Qing et al. (2006) evaluated the pressure fluctuations and vibrations caused by an orifice in a piping system. They suggested that the orifice induced substantial disturbances in the internal flow, amplifying pressure fluctuations, particularly in the low-frequency range, which may lead to resonance and structural fatigue in practical applications. Zhang et al. (2015) used CFD simulations with LES to study flow-induced noise generated by guide vanes in a 90° elbow pipe. Their findings showed that guide vanes effectively reduced turbulence and mitigated the sound pressure levels, thereby decreasing vibrations. Nevertheless, the study primarily addressed noise control in specific geometries, with limited focus on dynamic pressure fluctuations in straight pipes or under multiphase flow conditions. Nguyen et al. (2021) and Nguyen et al. (2023) conducted experimental analyses on the hydrodynamic characteristics of two-phase flow disturbed by globe valves and butterfly valves installed in piping systems. They also quantitatively assessed the altered pressure distribution in the piping system due to the valves. das Neves et al. (2024) experimentally studied the pressure signal generated by two-phase flow in horizontal pipes to identify flow patterns within the pipe. Chen et al. (2023) developed a method to identify the flow patterns in a gas–liquid pipe using a CNN (Convolution Neural Network) model based on flow-induced vibration data. Huang et al. (2025) examined the characteristics of intermittent flow in pipes using a visualization measurement technique.
Although previous studies have provided valuable insights into flow-induced pressure fluctuations, most studies have focused on the phenomena caused by fittings and components, such as orifices, bends, and valves. Consequently, pressure fluctuations in straight pipe sections, which are fundamental elements of piping systems, have not been investigated sufficiently. In addition, many studies rely heavily on CFD-based numerical simulations because of experimental limitations, leading to a lack of quantitative experimental data on the pressure fluctuations induced by multiphase flow in straight pipes. This lack of data hinders a fundamental understanding of pressure fluctuations that directly contribute to flow-induced loads, vibrations, and noise in piping systems.
This study quantitatively investigated the pressure fluctuations and their spatial distribution on pipe walls induced by single-phase and two-phase flows in a straight pipe. The experiments were conducted using a flow test loop to measure pressure fluctuations under various flow velocities and water fractions. The experimental results were also compared with a 1-D numerical model to examine the influence of the reservoir tank on the pressure magnitudes and distributions within the piping system. The pressure fluctuation characteristics were analyzed further in relation to free-surface deformation under different flow conditions. These findings provide experimental insight into the pressure fluctuations in straight pipes under multiphase flow conditions, contributing to a better understanding of flow-induced loads and vibrations in piping systems.
The paper is organized as follows. Chapter 2 provides a brief description of the experimental setup and conditions. Chapter 3 presents the preliminary studies of the flow characteristics under conditions where the piping system operates in either an open or closed loop, as well as the flow patterns in piping for two-phase flow. Chapter 4 reports an analysis of the pressure fluctuations under various experimental conditions, and Chapter 5 presents the conclusions.
2. Experimental Setup and Conditions
The experimental study was carried out using a two-phase (air-water) flow test loop at Pusan National University, as shown in Fig. 1. The flow loop consisted of an 18-meter-long stainless-steel pipeline with a nominal diameter of three inches. It included a four-meter-long acrylic test section (Fig. 1(c)) for the measurements and flow visualization. Fluid flow within the system was driven by a multi-stage centrifugal pump (Fig. 1(d)) with a maximum operating speed of 3,500 rpm. A reservoir tank (Fig. 1(e)) with a total volume of 2 m3 was connected to the pipeline to temporarily store water and release air from the fluid flow to the atmosphere.
Two-phase (air-water) flow test loop at Pusan National University: (a) Schematic diagram, (b) Full shot, (c) Test section, (d) Pump, and (e) Reservoir tank
Six pressure sensors were installed at regular intervals along the pipeline, both above and below, to measure the pressure distribution caused by fluid flow, as shown in Fig. 2. The sensors used in the experiments were Kistler 4043A2 piezoresistive models, capable of measuring the static and dynamic pressures within a range of 0 to 2 bar. The sampling rate of the pressure measurements was 20 kHz during the experiments, and a finite impulse response (FIR) band-pass filter with a cut-off frequency range of 0.7 to 2,000 Hz was applied to eliminate the background noise. Pressure data were collected for five minutes in all test cases, and the recorded data represented gauge pressure.
In addition, a vortex-type flow meter (KTV-700) was installed upstream of the test section to measure the flow rate within the loop. This flow meter, with a capacity of 2–10 m3 /h, provided the flow data at a sampling rate of 50 Hz.
The characteristics of the free surface in multiphase flow were analyzed by capturing instantaneous images of the free surface using a high-speed charge-coupled device (CCD) camera (Redlake Y-5) and an LED backlight, as shown in Fig. 3. The camera was equipped with a 50 mm f/1.8 macro lens set to f/4.0. Images were acquired at 100 Hz, covering a field of view (FOV) width of 0.66 m to observe free surface deformation within the pipeline.
The study examined five different flow speeds in the test loop, based on the operating range of the centrifugal pump. The flow speeds varied from 1.00 to 2.00 m/s, corresponding to Reynolds numbers (Re, calculated using Eq. (1)) ranging from 70,000 to 140,000, as listed in Table 1.
where UFs is flow speed of water, D is pipe diameter, ν is kinematic viscosity. The experiments also included variations in the water fraction, defined in this study as the volumetric ratio of water occupying the internal cross-section of the pipe. The water fraction was varied from 90% to 100% in 2% increments. A water fraction of 100% was achieved by allowing water to flow through the reservoir tank while releasing air from the pipeline. Some water was drained while isolating the pipeline from the reservoir tank to reduce the water fraction. In both cases, the pump speeds were kept constant, and the water level in the reservoir tank was maintained at 1.10 m.
3. Preliminary Study of the Flow Characteristics in a Pipe
Before analyzing the pressure fluctuations caused by multiphase flow in a horizontal pipe, preliminary research was conducted to investigate the pressure distribution characteristics and flow patterns within the flow test loop. This study focused on the single- and two-phase flow conditions, aiming to establish a baseline understanding of the pressure behavior and flow dynamics across various operating conditions in the flow test loop.
3.1 Effect of Reservoir Tank in Piping System
Fig. 4 shows the pressure distribution in the test section for a 100% water fraction under two conditions: an open loop with the reservoir tank connected and a closed loop without the reservoir tank. The pressure values represent five-minute averages, with the error bars indicating the standard deviations. The pressure was highest at points P1 and P4 near the inlet and gradually decreased along the pipe because of the major losses caused by friction on the pipe wall. The pressure also increased as the flow velocity increased. The pressures at the bottom (P1–P3) and top (P4–P6) of the pipe were nearly identical because the flow consisted entirely of water without air. In the open loop (Fig. 4(a)), however, all pressures acting on the pipe were positive, while they were negative in the closed loop, with a pressure difference of approximately 10–20 kPa. This difference likely results from the hydrostatic pressure of water stored in the reservoir tank in the open loop.
Comparison of the averaged pressure measured in the pipe between the open and closed loop flows (with and without connecting reservoir tank)
The difference in pressure distribution characteristics between open and closed conditions in the flow loop was analyzed by simulating the entire flow loop system using PIPENET. PIPENET is a 1D unsteady flow simulation program that solves mass, momentum, and energy conservation equations to predict the pressure, flow rate, and velocity profiles within pipe networks. It uses finite difference methods to discretize and solve the governing equations, enabling rapid analysis of the transient and steady-state flow behaviors in pipe networks. For more details about the program, please refer to Sunrise System Limited (2020).
The flow loop used in the experiment was numerically modeled in PIPENET (Fig. 5), and the estimated pressure distribution at the test section was compared with the experimental results in Fig. 6. Although quantitative discrepancies were observed between the experimental results and the simulation, the numerical model captured the overall pressure distribution trend along the piping system. These discrepancies may be attributed to the limitations of the PIPENET model, which is based on a one-dimensional averaged steady-flow formulation. Therefore, they cannot fully capture local flow disturbances, three-dimensional flow structures, and transient flow characteristics. In addition, pump disturbances, local losses caused by fittings and bends, and differences in boundary conditions between the open-loop and closed-loop systems may have contributed to the observed discrepancies.
Comparison of the experimental and numerical results of averaged pressure for open and closed loop flows (with and without reservoir tank)
Fig. 7 shows the pressure distribution along the entire flow loop system estimated using PIPENET. The pump location is designated as the starting point, with the test section highlighted by a red box. In the open-loop configuration (Fig. 7(a)), the pressure reached its maximum at the starting point because of the pump head, then gradually decreased along the piping due to major losses as the fluid flows, with greater losses at higher flow velocities. As the fluid passed through the vertical pipe entering the reservoir tank (≈16 m from the pump), the pressure dropped to negative values across all flow velocities. In contrast, the pressure decreased continuously in the closed-loop configuration (Fig. 7(b)), where the reservoir tank was bypassed, due to significant losses, resulting in negative pressure at the test section. This analysis highlights how the presence of a reservoir tank in the piping system can significantly influence the pressure characteristics, particularly the pressure magnitude, under different operating conditions.
3.2 Flow Pattern of Two-Phase Flow
In this study, the flow pattern of the two-phase flow was analyzed for different velocities and water fractions in the flow test loop. These patterns, referred to as flow regimes, vary with the velocities of liquid and air, exhibiting various flow structures. Fig. 8 shows the instantaneous images of the free surface obtained with a CCD camera while the water fraction was varied at a velocity of 2.00 m/s. At the smallest water fraction of 90%, the flow exhibited the characteristics of annular flow because of a large amount of air in the pipe, while the shape of slug flow was observed as air clusters at water fractions of 92% (Fig. 8(b)) and 94% (Fig. 8(c)). At water fractions of 96% and 98%, numerous small bubbles flowed along the free surface, showing the pattern of bubbly flow.
In this study, the water-fraction conditions at flow velocities of 1.00 m/s and 2.00 m/s are represented by symbols on the flow regime map (Weisman & Patterns, 1983), as shown in Fig. 9. The transitions between flow regimes highlight how the flow velocity and water fraction influence the flow structures. As the flow velocity increased and the water fraction decreased, the flow conditions moved toward the annular–slug transition region in the flow regime map. Conversely, increasing the water fraction or decreasing the flow velocity shifted the flow conditions toward the bubbly-flow region. These changes indicate that the interface structure and free-surface behavior differ within the pipe. At the flow velocities and water fractions considered, all conditions fell into the category of slug flow, gradually approaching bubbly flow as the water fraction increased or velocity decreased, and annular flow as the water fraction decreased or the velocity increased.
Flow regimes of two-phase flow for experimental conditions (adapted from Weisman and Patterns, 1983)
4. Results
The experimental results in this study are presented primarily in the frequency domain for analyzing pressure fluctuations, except for Fig. 10, which shows the time histories. The pressure fluctuations were analyzed by collecting data at 4 kHz during post-processing, considering the Nyquist frequency to avoid aliasing. For fast Fourier transform (FFT) analysis, a 1.25-second segmentation with 66.67% overlap and a Hanning window were applied. The FFT results were averaged for each segment. The results are shown as power spectral density (PSD), normalized by the hydrostatic pressure from the reservoir tank, and displayed on a logarithmic scale for both axes.
Comparison of the pressure time histories measured in experiments for open loop varying flow speed and measurement locations
4.1 Pressure Fluctuation by Single-phase (water) Flow
Fig. 10 shows the pressure distributions measured at various flow velocities and locations over time for single-phase flow under the open-loop condition. Fig. 10(a) presents the pressure variations at location P5 as flow velocity increases, showing higher magnitudes and greater fluctuations at higher velocities. The pressure variations at different locations were due solely to static pressure differences (Fig. 10(b)), with the fluctuation magnitudes and phases remaining consistent, suggesting that measurement location does not significantly affect the pressure characteristics. Overall, single-phase flow fluctuations may be influenced primarily by the flow velocity rather than the measurement position, underscoring the uniform behavior of the pressure fluctuations throughout the system.
Fig. 11 shows the normalized PSD of the data in Fig. 10 in the frequency domain. The pressure fluctuations showed a first peak around 10 Hz and a second peak around 20–30 Hz. The first peak, ranging from 0.001–0.1 Hz−1, likely results from water descending from the upper part of the reservoir tank to the free surface. The second peak appears to correspond to the pump rotation speed, influenced by the flow velocity (Fig. 11(a)). Additional peaks at higher frequencies may be due to mechanical equipment in the loop. Fig. 11(b) indicates similar magnitudes and frequencies of pressure fluctuations across measurement locations. This result is consistent with Fig. 10(b), which also showed uniform pressure fluctuation characteristics across different measurement points. These frequency components may act as excitation sources for flow-induced vibration and noise in piping systems.
Comparison of the pressures in the frequency domain for open-loop varying flow speed and measurement locations
Fig. 12 presents the pressure fluctuations for closed-loop conditions with a 100% water fraction. A notable finding is the absence of the first peak at around 10Hz, which was evident in the open loop, suggesting reduced fluctuations caused by the reservoir tank as the fluid bypassed the tank. Other frequency patterns between the open and closed loops remain similar.
To assess the magnitude of the sound level from the fluid-induced pressure fluctuation, the pressure fluctuations measured in the experiment were compared the octave-band sound power level from flow-generated noise (Lw, OCT) using Ver and Beranek (2005) formulas (Eq. (2)).
where P is atmospheric pressure; c is the speed of sound; S is the cross-sectional area of the duct; Ma is the Mach number; H is the maximum dimension of geometry; δ is the length scale characterizing the high-frequency spectral content of regenerated noise (in this study, H and δ represent the diameter of the pipe); and ν is the velocity. B is a coefficient that varies depending on the shape (95 in the study). The pressure fluctuations were converted to equivalent sound pressure (Pflow) using Eq. (3),
Fig. 13 compares the experimental results with the estimated sound level. The analytical model showed good agreement with the experimental results within the frequency range of 10–500 Hz. On the other hand, some discrepancies were observed below 10 Hz and above 500 Hz. The larger fluctuations in the low-frequency range are considered to be associated with free-surface oscillations in the reservoir tank. In contrast, the discrepancies in the high-frequency range may be related to local turbulent structures and pump-induced disturbances that are not accounted for in Eq. (2). Nevertheless, this approach effectively estimates the noise levels within the 10–500 Hz range. It provides practical guidance for optimizing fluid system designs and addressing noise control challenges.
Comparison of the pressures for open loop with analytical flow-generated noise suggested by Ver and Beranek (2005)
4.2 Pressure Fluctuation by Two-phase Flow
The pressure fluctuations caused by two-phase flow were analyzed using image processing techniques applied to high-speed CCD camera images. The free surface in each image was identified in MATLAB using edge detection and image filtering methods to analyze the fluctuation frequency, as shown in Fig. 14. Fig. 15 presents the dominant free-surface fluctuation frequencies extracted from the image-processing analysis. Across the conditions, the fluctuation frequencies did not exceed 2 Hz, with the highest peak observed at a 98% water fraction due to the small bubble formations. Except for the case of a water fraction of 98%, the fluctuation frequency of the free surface did not show a clear trend within the experimental conditions of flow velocity and water fraction. The dominant fluctuation frequencies extracted from the image-processing analysis were compared with the pressure fluctuation characteristics discussed in the following section.
Fig. 16 compares the pressure fluctuations for varying water fractions at a flow velocity of 2.00 m/s. Unlike single-phase flow, two-phase flow showed increased fluctuation magnitudes at low frequencies below 10 Hz. The first peak increased as the water fraction decreased, reflecting the interactions between the free surface and the pipe because of air. This tendency is consistent with the flow regime characteristics shown in Fig. 9, where lower water fractions exhibited stronger interface deformation and flow characteristics closer to the annular–slug transition region. The increase in low-frequency pressure fluctuations at lower water fractions is also consistent with the enhanced free-surface fluctuation observed in the image-processing analysis (Fig. 15). Moreover, unlike the single-phase flow, the second peak caused by the pump was absent, which can be attributed to the decrease in pressure oscillations induced by the pump, likely dampened by the compressibility of air within the pipe.
Fig. 17 compares the characteristics of the pressure fluctuations for various flow velocities and measurement locations at a 98% water fraction. As shown in Fig. 17(a), in two-phase flow, even with an increased flow velocity, the overall pressure fluctuation within the piping did not change significantly. This confirms that the flow velocity did not significantly affect the fluctuations in two-phase flow, unlike single-phase flow. In the frequency range below10 Hz, the normalized PSD varied by approximately 0.01–1 Hz−1 depending on the longitudinal position within the pipe (Fig. 187b). In the 10–200 Hz frequency range, higher pressure fluctuations were observed at the upper part of the pipe where it was directly affected by the free surface. In the high-frequency band above 200 Hz, the fluctuating pressure measured below the piping was greater than that measured at the upper part, which might be due to the structural characteristics of the flow loop supported from below.
5. Conclusions
This study experimentally investigated the characteristics and distributions of the pressure fluctuations induced by single-phase and two-phase flows in a horizontal straight pipe system. The experiments were conducted in a flow test loop to quantify the pressure fluctuations and free-surface deformations under varying flow velocities and water fractions. The key findings are summarized below:
(1) The experimental and numerical results showed that the pressure magnitudes within a piping system vary significantly depending on whether the system operates in an open or closed loop configuration. The open-loop conditions resulted in higher pressures from atmospheric exposure, whereas closed-loop conditions exhibited negative pressures throughout the piping system because of the absence of hydrostatic effects.
(2) Significant pressure fluctuations were observed below 10 Hz during single-phase flow, primarily influenced by the free surface effect of the reservoir tank. These fluctuations diminished under closed-loop conditions. In addition, within the 10–500 Hz frequency range, the pressure fluctuations closely matched the predictions based on the flow-induced noise formula proposed by Ver and Beranek (2005), validating its applicability for practical designs.
(3) In two-phase flow, the pressure fluctuations increased in the low-frequency range below 10 Hz compared to single-phase flow due to the presence of air. The magnitude of the fluctuations increased as the water fraction decreased, indicating the influence of severe free surface deformation and interface instability. Furthermore, pressure oscillations caused by the pump and other piping equipment were reduced under two-phase flow conditions, which is presumed to be related to the damping effect of air within the pipe.
The results provide fundamental experimental data for understanding the pressure fluctuations induced by internal flow in straight piping systems. Such pressure fluctuations can act as excitation sources, leading to flow-induced vibration and flow-generated noise in piping networks. In marine and offshore engineering systems, piping networks are used widely in ships, submarines, and offshore platforms to transport fluids such as cooling water, fuel, and ballast water. Therefore, understanding the characteristics of flow-induced pressure fluctuations is essential for evaluating structural vibrations, noise radiation, and structural integrity of such systems. The experimental data and observations presented in this study provide fundamental reference data for the analysis and prediction of flow-induced excitation in marine piping systems. Future studies should extend the present work to more complex piping geometries and components commonly used in marine systems, such as elbows, valves, and pumps.
Notes
The authors declare no potential conflict of interest.
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS-2024-00405278 and RS-2025-21063039), Korea Institute of Marine Science & Technology Promotion (KIMST) funded by the Ministry of Oceans and Fisheries (RS-2025-02219107), Korea Institute of Energy Technology Evaluation and Planning (KETEP) and the Ministry of Climate, Energy & Environment (MCEE) of the Republic of Korea. (No. RS-2026-25537760)
