(Editor's Note: There is a figure that accompanies this article that can be downloaded using the "Dowload Now" button at the bottom of the page.)
Some problems have no ideal solutions. The troubleshooter instead must identify the best compromise between the desired objectives. The following case illustrates such a situation.
Repeated attempts to use a conventional centrifugal pump for handling compressor interstage liquids failed. The compressor recovered fuel from a flare-gas system by condensing light hydrocarbons in the interstage coolers and aftercoolers. The compressor suction operated at around 1 psig, the interstage at 19 psig and the discharge at 160 psig. The original intent was to send the recovered liquids to other units. However, the pump suffered from repeated failures due to low available net positive suction head (NPSH
As a temporary solution, liquids were recycled from the interstage and aftercooler drums to the suction drum by pressure. This created several problems: Liquid recycle built up during operation, gas recovery from the flare system was reduced, and compressor reliability problems were created. So, a project was started to find alternative destinations for the liquid streams.
Three possible downstream locations were identified, but none would work at every operating condition. The pressures at these destinations were 190 psig, 230 psig and 290 psig; to reliably send the liquid to any of these required extra equipment.
The root cause of the problem was the original compressor installation. The exchangers and drums were built too low. Pump selection can never solve a poor design. However, when you have to live with a situation, you make the best choice you can.
Conceptual process engineering identified new pumps as necessary. Piston pumps with variable-speed motors appeared to be the only choice to meet the constraints of discharge head, available suction head, and flow-rate flexibility. The interstage drum and the aftercooler drum would each have separate pumps, and variable-speed drives would control the flow rate. The piston pumps would be able to deal with the full flow-rate range, composition variations and low suction head. Essentially, this duplicated the intent of the original process design, but would work. However, appropriate control systems for the two pumps were very expensive in both capital and maintenance.
So, alternative options to reduce the investment and operating cost were reviewed and a list was made of why they were discarded. From this list, one key factor stood out. The major constraint that eliminated many other choices was the need to have a 271-psi differential pressure rise for the interstage drum liquids (19-psig suction to 290-psig discharge). If that limit could be relaxed, then much cheaper solutions were available.
When faced with difficult equipment constraints, never forget to reexamine the process. Process and equipment choices interact; equipment is not independent of the process. Modifying the liquids handling (Figure) by routing the interstage liquids to the aftercooler drum reduces the head required for the first stage liquids from 271 psig to 141 psig. Sliding vane pumps are a practical choice for a 141-psig rise as they work well in a differential pressure range of 160-180 psi, maximum. The new configuration uses sliding vane pumps with variable-speed drivers. The result: Capital was reduced by 50% and maintenance costs were lowered as well.
The sliding vane pump with a variable-speed driver is not ideal; but no pump is perfect for this service. This solution delivers an acceptable combination of results at a lower cost.