Aug-2016

Loss into gain in high-capacity trays Part 2: reverse vapour cross flow chanelling

Systematic troubleshooting diagnosed two complex problems that limited the capacity of an atmospheric crude distillation tower

HENRY KISTER, Fluor
NEASAN O’SHEA (ret’d) and DAN CRONIN, Phillips 66 Whitegate refinery

Article Summary

In the 1990s, Irish Refining Company operated a 52 500 b/d (7070 t/d) crude unit processing North Sea crudes (this refinery is now owned by Phillips 66). A detailed description of the crude tower and products is in Part 1 of this article (PTQ, Q2 2016).

Initially, the four-metre ID atmospheric crude tower contained 34 single-pass jet tab trays. At a later time, the tower was revamped by others to increase fractionation efficiency and to permit the refinery to intermittently take a kerosene side-cut while maintaining the same throughput of 52500 b/d (7070 t/d). Trays 11-20, the trays in the heavy naphtha (HN) diesel fractionation section, were replaced by single-pass, high-capacity Nye trays. Structured packings were installed in the wash and stripping sections. A new kerosene side draw was installed at the inlet to Tray 16, approximately mid-way between the HN and diesel draws. A side stripper was added to strip the kerosene. Initially, the kerosene rundown system was incomplete so the kerosene side draw was not operated.

After the retray, the tower was not able to achieve the 52 500 b/d (7070 t/d) it had previously achieved. At throughputs exceeding 38 000 b/d (about 5200 t/d), the tower flooded with a rise in pressure drop and the HN product going off-specification with heavy components. Initial gamma scans showed the flood initiated near Trays 14-16, building up until Tray 23. The pumparound Trays 21-23 were jet tab trays and were not changed in the revamp.

In addition, a severe instability in the lower part of the column, which we termed an ‘excursion’, would usually occur at about the same time and rates as the tray flooding was initiated. This excursion was caused by the diesel draw nozzle and rundown line running into a self-venting flow limitation following a revamp modification that reduced the liquid residence time of the diesel draw pan. The diagnosis and correction of this problem are described in detail in Part 1 of this article. This Part 2 of our article focuses on troubleshooting and correcting the cause of the premature flood.

Channelling to explain the premature flood
Figure 1 shows the tower with the post-revamp mass balance and operating conditions at a reduced crude rate of 38 700 b/d (5200 t/d) in black print. Also shown on the diagram, in red print, are the mass balance and operating conditions following the operational debottleneck to 49100 b/d achieved by taking a kerosene draw that will be discussed later in this article.

Trays 11-20 are high-capacity Nye trays² (see Figure 2). These trays differ from conventional trays by having a ‘Nye unit’ at the tray floor, right under the downcomer. The Nye unit is closed at the top, perforated at the side and underneath. Vapour entering the unit from the tray below via the bottom holes is turned horizontally towards the tray and exits via the side holes. This horizontal wind imparts a horizontal ‘push’, or horizontal momentum, to liquid droplets in the inter-tray space, blowing them towards the outlet downcomer and preventing their entrainment into the tray above. By reducing this entrainment, the horizontal push is the capacity enhancement mechanism on the Nye tray. The same horizontal push mechanism, produced by a variety of devices and geometries, is employed by most of today’s high-capacity trays.

Figure 3 (roughly to scale) shows the tray dimensions. The active areas of Trays 11-16 contained long-legged venturi (smooth orifice) uncaged moving valves with 16.2% open area (the most restrictive open area of the fully open valve, expressed as a percent of active area), while Trays 17-19 contained small-diameter sharp-orifice moving valves with 14.6% open area. There were also small differences in downcomer and active area dimensions (see Figure 3).

Hydraulic calculations both by the vendor and the troubleshooting team (see Table 1, the column headed ‘Before operational debottleneck’) showed that at 38 700 b/d crude charge, the most highly loaded trays above the diesel draw and below the top pumparound (Trays 11-20) should have been operating comfortably below any capacity limit. The calculations showed that at these loads even conventional trays should have operated without flood.

Eye-catchers in Figure 3 and Table 1 are the very long flow paths, giving a huge ratio (3.6-4) of flow path length to tray spacing, the very large weir loads (100-116 m3/h m of outlet weir), the venturi valves, and the large open areas (14.6-16.2%). Previous work showed that venturi valves, large open areas (>13-14% with sharp orifice moving valve trays), large ratios of flow path length to tray spacing (> 2:1), and high weir loads (>50-60 m3/h m of outlet weir) render trays prone to vapour cross flow channelling (VCFC).^3,4 VCFC is encountered when, under the 
influence of a hydraulic gradient, vapour preferentially channels through the tray outlet and middle, generating a high vapour velocity region with high entrainment and premature flood. At the same time, the tray inlet region remains vapour deficient, which promotes excessive weeping and poor tray efficiency.

VCFC is not the only form of channelling experienced on distillation trays. There are reports^5,6 of other forms of channelling, such as due to vapour maldistribution or multi-pass tray maldistribution. One thing they have in common is that they have only been experienced at large tray open areas, high ratios of flow path lengths to tray spacing, and high weir loads, conditions that apply also to the trays involved in this case.

An argument against VCFC is that forward push was used as a cure to VCFC. In VCFC, the channelling is caused by an excessive hydraulic gradient.^3,4,7 Using ‘push valves’, slots or valves that enter some of the vapour onto the tray with a horizontal velocity in the direction of the liquid flow, can reduce the hydraulic gradient and alleviate VCFC⁸ as has been demonstrated by one published case study⁹ and some others experienced by the authors. The Nye trays used in this tower entered about 10-15% of the hole area in a horizontal direction, providing considerable forward push. This forward push should have been sufficient to counter the hydraulic gradient.

It therefore became necessary to seek further insight into the froth pattern on Trays 11-20. Channelling is difficult, often impossible, to diagnose with conventional troubleshooting techniques such as vendor software, ∆P measurement, and conventional single-chord qualitative gamma scans. Judicious multi-chordal gamma scans with quantitative analysis is the best tool for diagnosing maldistribution on trays. The principles and practice of this technique were recently described in detail elsewhere.¹ This technique was applied here to give a concise definition of the trays‘ hydraulics and to diagnose the root cause of the premature flood.

Multichordal gamma scans with quantitative analysis
Four gamma scan chords were shot through the active areas of Trays 9-22 parallel to the downcomers. In addition, two gamma scan chords were shot perpendicular to the downcomers. These perpendicular chords were intended to pass through the active areas only, and to miss the downcomers. All chords were shot when the tower was just below flood at 37500 b/d charge rate (‘incipient flood’) and repeated when the tower was flooded at 44 300 b/d charge rate. During the scans, column operation was kept extremely stable and steady. This was vital as for the scans to be consistent with one another no change could be allowed to the process during the shooting period. At only one time instability was experienced for a short time, and the data obtained during this instability were discarded.