Salem Bouhairie, Group Lead, Research

Air-cooled heat exchangers (ACHEs) in the oil, gas, and power industries often process two-phase fluids, and their performance is usually limited by the airside thermal transfer resistance. However, flow on the tube side can undergo phase change, such as by vapor condensation or liquid vaporization, producing complex two-phase behavior.

Complications from thermal and hydraulic maldistribution within the header can induce uneven heat transfer between airside and tubeside flows. Two-phase maldistribution in headers negatively affects the output products and can cause tube pullout. Maldistribution can also affect the mechanical integrity of the ACHE where piping becomes a safety issue if pressures are unfavorable.

In a conventional API-661 [1] ACHE, the process fluid enters one or more inlet nozzles in the distributing header (or inlet header) and flows through the tubes. Figure 1 shows a two-pass ACHE, where the process fluid flows between the front and rear headers as the stream progresses from one pass to the next. After the last pass, the flow enters the collecting header (or outlet header) and into one or more outlet nozzles.

Figure 1. Process flow path from inlet header to outlet header
Figure 1. Process flow path from inlet header to outlet header

While the distributing and collecting headers of an air-cooled heat exchanger are typically rectangular (i.e., boxes), the cross section may be circular or semicircular to accommodate high pressures or other process considerations. Designers need to better understand how the header distributes or collects the flow and how to calculate the header pressure drop in order to obtain a consistent tubeside flow that allows uniform cooling throughout the heat exchanger.

Phase separation in headers

One important regime of two-phase maldistribution is phase separation, which challenges the accepted design approach and negatively affects the operation of ACHEs. Open literature has few studies on phase separation of pure components and mixtures in API [1] ACHE headers. Xace®—HTRI's software for designing, rating, and simulating air coolers and economizers—predicts performance loss if phase separation occurs in ACHE headers, and additional improvements to its correlations are in process.

API recommendations

Two-phase flow dynamics are complicated and notoriously difficult to predict. According to API 661 [1], two-phase fluids in a single pass with multiple rows require a more complete analysis that recognizes the separation of phases in the header. API 661 [1] defines criteria to prevent flow maldistribution inside air-cooler headers, but these criteria are very conservative and usually result in unnecessary increases of manufacturing costs. In addition, API 661 does not establish a procedure for estimating the extent of flow maldistribution [2].

Xace single-phase maldistribution

HTRI's modeling of header maldistribution is conducted in stages. Xace 8.0 accounts for pressure loss in the header and reports single-phase maldistribution for liquid services only. AC-18 [2] details how the magnitude of the single-phase flow maldistribution in the tubes of an ACHE can be expressed in terms of relative standard deviation (RSD) and the maximum mass flow rate variation (ΔV+). The RSD is the normalized standard deviation in the mass flow rate between the tubes, expressed as a percentage. The maximum variation in the mass flow rate, ΔV+, is the normalized difference of the maximum and average flow rates inside the tubes. These two variables quantify deviation from uniform distribution. Flow maldistribution could be considered good if below 5%, satisfactory if between 5% and 10% , and poor if above 10% [2].

Experimental testing of two-phase maldistribution

To improve our methods, HTRI designed and built a scaled-down model of the header and tube bundle for adiabatic air-water testing in the Multipurpose Visualization Unit (MVU). Using the MVU test data, we can propose recommendations for effective header design and later modify Xace to report a two-phase header flow maldistribution parameter [2].

Tests on the small-scale MVU with eight tubes demonstrated that we could reasonably visualize two-phase flow maldistribution. We next built a larger MVU test section with 120 tubes to collect more industrially relevant data. Figure 2 shows the two scales of the MVU. Table 1 summarizes the geometry and process conditions of the full-scale MVU.

Figure 2. Proof-of-concept (left) and full-scale MVU for lateral maldistribution studies (right)
Figure 2. Proof-of-concept (left) and full-scale MVU for lateral maldistribution studies (right)

Table I. Full-scale MVU test unit geomery and process conditions

Parameter Value
Tubes / rows / passes 120 / 4 / 1
Tube OD / ID, mm 25.4 / 19
Tube length / bundle width, m 2.4 / 1.95
Tube pitch transverse / longitudinal, mm 64 / 55
Tubeside temperature in / out, °C 22 – 35 / 22 – 35 (adiabatic)
Airside temperature in / out, °C 22 – 35 / 22 – 35 (adiabatic)
Water / air flow rates, kg/s 1.3 – 10.7 / 0.025 – 0.12

The videos below show MVU tests with water filling the header and pushing out air as slugs. These visualizations provide initial insight into two-phase flow behavior.

Videos: Air slugs are pushed out through tubes in initial tests with water in MVU

The graph in Figure 3 plots the single-phase water measurements in the header at certain tube columns (indicated in the sketch by the staggered red dots). Discrete measurements were taken at every seventh tube, to provide detailed local flow and pressure drop distributions vertically and laterally. At the inlet nozzle (tube column 3), peak flow occurs at the lowest tuberow; minimum water flow occurs in the topmost tuberow. Lateral-most side tubes (columns 1 and 5) experience local water flow maxima.

Figure 3. Lateral distribution of single-phase water flow measurements shows peaks at nozzle in lowest row and local maxima at sides
Figure 3. Lateral distribution of single-phase water flow measurements shows peaks at nozzle in lowest row and local maxima at sides

Future plans

Modifying MVU header partitions will enable testing in a wider range of air-water mixture flow regimes. Using these MVU data, HTRI plans to improve our forced phase-separation methods and recommend changes to header designs that can reduce the negative impact of phase maldistribution on ACHE performance.

Visit this webpage to learn more about our air-cooler research.

References

  1. API STD 661: Petroleum, Petrochemical, and Natural Gas Industries—Air-cooled Heat Exchangers, 7th ed., American Petroleum Institute, Washington, DC (2013).
  2. M. Rezasoltani, Air-cooler header pressure drop and flow maldistribution, AC-18, Heat Transfer Research, Inc., Navasota, TX (2017).