February 2019

Special Focus: Materials Manufacturing

Welded plate heat exchangers cut refinery process costs—Part 1

Plate and frame (P&F) heat exchangers were first manufactured 100 yr ago for use in the dairy industry.

Plate and frame (P&F) heat exchangers were first manufactured 100 yr ago for use in the dairy industry. They have since been used in the process industries, as they offer considerable advantages in true counter-current flow, high heat transfer rates, low manufacturing costs and a small footprint. These benefits lead to lower capital and operating costs than can be achieved with shell-and-tube (S&T) heat exchangers. The limitation on their use has been the elastomer gasket between each plate, which is susceptible to leakage from chemical attack or to penetration by aromatics. In addition, operating pressure and temperature are typically limited to 20 bar at 180°C (356°F) for water, conditions that preclude its use in many downstream processes, especially refining, where aromatics can soften gasket materials at considerably lower temperatures.

FIG. 1. Typical plate pack arrangement.
FIG. 1. Typical plate pack arrangement.

Welded plate exchangers arose from these shortcomings in the P&F type. Two types exist: plate-shell and block. Both were initially tungsten inert gas (TIG) welded, but are now fabricated with laser-welded plates. Production of the laser-welded plate-shell type started in 2006, and some 1,200 units have since been delivered, mainly for use in the chemical industry or for utility systems in large buildings. The plate-shell type differs from the block type in two vital respects: it has true counter-current flow, whereas the block type is a cross-flow design, and the service pressure for the plate-shell heat exchanger is potentially as high as 400 barg, comparable to the S&T exchanger, whereas the block type is limited to around 30 barg by its cuboid shape.


The construction of a plate-shell heat exchanger is shown in FIG. 1. The plate pack is cylindrical, with circular plates typically 0.8 mm thick, depending on the design pressure and temperature, pressed with a pattern approximately 3 mm deep, leading to an area of 300 m2/m3. An S&T’s area is less than 100 m2/m3. As the plate heat exchanger typically has a U-value at least double that for the S&T equivalent, a six-fold or more reduction in size can be achieved, and the unit is less prone to fouling than the S&T type, as there is much higher wall shear stress arising from the turbulence created by the plate pattern.1 This compactness, combined with the low fouling rate and the inherent strength of a laser-welded plate pack (the weld has never failed during burst tests), leads to significant possibilities for its use in refining and petrochemical processes. The main components are the plate pack assembly; the cylindrical shell, which is little more than a conventional pressure vessel; and the flow directors, which are of flexible stainless-steel construction.

Plate packs

The plate pack is assembled between thick-walled retaining plates. The plates are pressed from 316-grade stainless steel or better (e.g., titanium, 254SMO, SS904L or Hastelloy), with accurate spacing being obtained by the high points on adjacent ribbed plates. Alternate plates are laser-welded at the rim to create the “shell-side” passages, and the holes in the plate are for the “plate-side” passages. The shell-side passages are open to the annulus formed between the plate pack and the shell. The plate-side passages are connected by inlet and outlet headers set at the top and bottom of the plates. The diameter of these connectors—and of the plate-side connections on the front of the shell—are geometrically limited so that too much of the available plate area is not lost. The effective plate area between these connections is approximately two thirds of the total circular plate area.

High theta plates with an obtuse angle for the ribbings, which increases the heat transfer rate and pressure drop, are used where a close temperature approach is needed. Number of transfer unit (NTU) values up to 3 are achieved in a single pass (e.g., a temperature cross is achievable in a single shell, something that is impossible with a two-pass U-tube or floating-head S&T heat exchanger). Low theta plates with an acute angle for the ribbings have a reduced heat transfer rate and lower pressure drop, and are ideal for condensers and reboilers, or where NTU values do not exceed 1.3 in a single pass. Flow deflectors may be used to make more than one pass within a single shell.

Flow directors

Flow directors (or bypass baffles) are vital; otherwise, the shell-side medium would take the path of least resistance and flow through the annulus, which is used for header space and distribution, rather than through the plates. It is possible to use elastomeric baffles, which fill the gaps between the plate pack and shell, or metallic baffles, which may not seal so well, but are resistant to chemical attack. Such metallic bypass baffles must be the same material as the plate pack to avoid any galvanic corrosion risk. Generally, metallic flow directors are preferred, as they are more robust.


The principal methods of application are like conventional S&T heat exchangers. As the shell diameter is comparable to the equivalent S&T exchanger, it is possible to retrofit an existing heat exchanger shell with a plate pack, either for debottlenecking reasons or to solve a corrosion problem.

Condensers and partial condensers. The natural solution is to put the condensing vapor in at the shell top and withdraw condensate from the bottom nozzle. Coolant circulates through the plate-side connections. It can be advantageous to increase the cooling medium film heat transfer coefficient by utilizing multiple passes on the plate side to increase the pressure drop.

It is also possible to use the plate pack as a stab-in condenser and to mount it at the top of the column above a chimney tray used for liquid draw-off, which is very compact and greatly reduces the total capital expenditure.

Reboilers. A kettle and a once-through thermosyphon reboiler are both possible, depending on process requirements.

For the kettle type, the liquid entering the bottom of the shell is partially vaporized and separated in the space above the plate pack, with the vapor exiting through the nozzle on top of the shell. The liquid, having a higher density than the homogeneous density of the two-phase fluid within the plate pack, recirculates through the annulus on the outside of the plate pack, as this design has no bypass baffle. The bottoms product is taken out over a weir, as in any other kettle reboiler. As the volume is much less than a traditional S&T kettle, startup and control are both faster. The much lower residence time is a key consideration for heat-sensitive products, such as amines, or when byproducts can be formed, such as heat-stable salts.

With a once-through reboiler, both the liquid and the vapor leave the reboiler and go back to the bottom of the column via the riser. This results in a smaller and less expensive shell than a conventional S&T exchanger. The once-through reboiler is also well suited to heat-sensitive products, owing to the low residence time. It is also more able to cope with movement, such as on a vessel (e.g., a floating production, storage and offloading vessel, or floating LNG vessel), as there is minimal sloshing.

For both types of reboiler, the heating medium is usually on the plate side, and a multiple pass arrangement can be utilized to improve the hot film heat transfer coefficient if the heating medium is a heat transfer fluid or hot water. A stab-in reboiler is also possible, which eliminates the cost of the shell and associated pipework.

Single-phase heat transfer. Unlike an S&T heat exchanger, there is no significant hydraulic difference between the two sides of a plate-shell type. The choice of which side to place each stream is more usually determined by maintenance considerations, with the cleanest fluid being sent to the plate side. If both fluids are clean, the lowest flow should be on the plate side. Given the compactness of the design, the advantage in placing a stream with a much higher design pressure on the plate side, rather than on the shell side, is much less of a consideration than for S&T heat exchangers.

Reliability and maintenance

Much testing was performed on the laser-welded plate-shell heat exchanger, including multiple burst tests on each plate’s material, thickness and size to ascertain the achievable design pressure. The integrity of the welding was proven with no failures of any welds, the only failures being with the parent material.

The next stage was to test the plate packs for thermal transients—as this is a common issue with all heat exchangers—during startup, shutdown and process upsets. Thermal cycling was performed between ambient temperature and –168°C (–270.4°F) at 80 barg, where in excess of 8,000 cycles were achieved before any failure occurred. Then, 36,000 cycles were performed between ambient temperature and 100°C (212°F). When the test was stopped, and a burst performed, the plate pack achieved the same results as a new plate pack.

More than 1,200 laser-welded, plate-shell heat exchangers have been installed, most of which have been placed in service within 12 mos. of delivery, in plants that are normally in continuous operation. This gives an accumulated operating experience of more than 50 MM shell-hours. Not a single plate-pack laser-weld failure has been reported since inception. Since the early years of manufacturing, the number of weld failures at the plate-size nozzle connection has been reduced to nearly zero, with mean time to failure in the order of 10 MM service hours.

In a few instances (representing less than 0.5% of all items), bypassing of flow has been reported around the flow deflectors, leading to some loss in thermal performance. Replacing the flow deflectors has helped resolve most of these problems.

Maintenance. The plate-shell heat exchanger handles fouling streams well, if it is designed with appropriate wall shear stress. The main reason is that the surface is akin to a static mixer, with the boundary layer constantly being broken to keep particles in solution. This also reduces the formation of scaling. Studies of fouling are in line with operating experience, in that the fouling is indeed slower and less severe than in an S&T exchanger. A full discussion of fouling mechanisms and their treatment is discussed in literature.2

On the principle that prevention is better than a cure, fitting a strainer, backflush connections or an in-situ flush system might all be considered to minimize downtime from fouling:

  • A strainer is needed if particles are larger than 1,000 microns (1 mm) (e.g., one third of the pressing depth).
  • A backflush connection on the first shell in a heat exchanger train can mitigate fouling and maximize run time. A differential pressure indicator is all that is needed to alert the operator to backflush a unit.
  • It is normally safe and economical to practice chemical cleaning in place (CIP), as the volume of fluid from plate-shell heat exchangers is small. This can easily be done in situ to avoid dismantling equipment. Kerosine might be chosen as a solvent for crude oil and residue streams. The dirty solvent can be sent to the slops system and re-run. A weak acid
    is needed for inorganic scaling in cooling water systems.

Due to the additional risk to health and safety, dismantling and jet washing should be considered only as a last resort. Given the limited access to the plate side of the unit, the normal preference would be to place (distillate) rundown and pumparound streams on the plate side, and to place potentially fouling streams, such as crude oil or residue streams, on the shell side, with a circulating wash needed only for the shell side.

Process applications

Following are process applications for plate-shell heat exchangers.

Revamps. As with any compact heat exchanger, a key advantage is that considerable space savings are possible over conventional S&T designs, leading to potentially significant economies in the cost of both new and revamp projects. Revamps are a low-risk and low-cost opportunity for testing out plate packs within existing equipment and processes to gain confidence in the equipment.

Refining. Studies have been carried out for potential use in crude distillation and product fractionation systems. The inherent strength and versatility of the plate pack should make it a natural choice for both thermal processes, which tend to be low pressure, as well as for hydroprocessing, as design pressure is not limited.

Petrochemicals. The application of plate-shell heat exchangers could be considered in ethylene plant cracked-gas compression, fractionation and refrigeration systems, where high-pressure streams under a range of operating temperatures are well within the capabilities of the laser-welded, stainless-steel plate pack.

Chemicals. A tube bundle was replaced on a hydrogen/methyl ester mixture interchanging with a hydrogen/fatty acid mixture at 200 barg–300 barg with a temperature profile of 20°C–300°C (68°F–572°F) at a major European chemical producer. The results were 250% of the area and a 400% increase in heat recovery, resulting in an 80% steam saving.

Gas treating. Studies of amine units (both diethanolamine and methyl diethanolamine) have shown significant benefits in using plate-shell heat exchangers in place of S&T heat exchangers, with as much as a 15%–20% reduction in total installed cost of the process unit.2,3 Application to amine streams in Claus tail gas units could also be considered.

Gas processing. Plate-shell units cannot replace multi-stream LNG-type heat exchangers, as they are limited to two streams, but could be considered where operating pressures are too high for aluminum plate-fin heat exchangers, or where streams are potentially corrosive to aluminum (e.g., gas from mercury-containing reservoirs).

NGL recovery and fractionation processes, including refrigerant systems, are possible uses for plate-shell heat exchangers, as the stream characteristics—clean and high pressure—are an ideal application of the equipment. Gasification and syngas processes are potential applications to consider, as well.

Part 2

Part 2 of this article, to be featured in the March issue of Hydrocarbon Processing, will present study results of a welded plate-shell heat exchanger in a crude distillation unit. The possibility of using a parallel scheme rather than a series flow heat exchange scheme for the crude preheat train is also investigated for both heat exchanger types. HP


  1. Bott, T. R., Fouling of Heat Exchangers, Elsevier Science, 1995.
  2. Broad, R. and P. Kauders, “Using welded plate heat exchangers to improve process economics,” Laurance Reid Gas Conditioning Conference, 2018.
  3. Broad, R. and P. Kauders, “Welded plate heat exchangers improve plant economics,” Gas Processors Association, 26th Annual Technical Conference, Muscat, Oman, March 2018.

Related Articles

From the Archive