Twist or STICK

Build-up of solid matter within process piping is a common problem in chemical plants. The problem is caused by solid materials, entrained in the gas flow, striking the pipe wall and sticking to it.

The problem of solids deposition is a function of the properties of the fluid carrier, flow rate, geometry of the pipe, and the size and weight of the entrained particles. Ideally, the fluid and entrained particles should follow smooth streamlines which track the geometry of the piping system. Under certain situations, however, the particles do not follow the fluid streamlines and instead hit the piping wall. This is more pronounced with tight bends, high flow rates and large particles.

Recently, a plant using Texaco-licensed technology was having severe build-up problems. Material was quickly building up on particular regions of the piping wall and reducing the throughput of the process. This would have eventually blocked the pipe, so this section required frequent cleaning. Because the problem area was at the front end of the process, this entailed the shutdown of the entire line, which cut revenues.

The problem was occurring so frequently, and the expense was so enormous, that we initiated a study to redesign the system. Historically, this type of problem was caused by the usual mechanism of high particle momentum. Thus it was tempting to simply decrease the throughput, increase the pipe radius or make some other empiricism-based change. However, we decided to use the Fluent computational fluid dynamics (CFD) software package to thoroughly understand the problem and design the most appropriate solution.

Initially, the plant's management was sceptical about the ability of CFD to accurately predict the complex flow conditions within the pipe; but, because of the costs they were incurring, they agreed to give it a try.

The first step was setting up a grid to model the flow within the pipe. The grid was generated so that its density was highest in the areas near the walls in order to properly simulate the turbulent boundary layer. The boundary conditions were selected so that if a particle hit a wall, it would stick to it.

Because the piping in the area of interest had a complex geometry, creating the model and entering boundary conditions took about a week. Since the model was created, an automatic mesher has been made available for use with Fluent that would probably reduce this time to about a day.

The computation itself took about three days to run on a Convex C210 superminicomputer. Again, advances in computer technology and CFD algorithms would now probably reduce this to less than a day.

The final solution was a great surprise. We expected to verify that high particle momentum was the primary cause of the build-up, but instead the CFD simulation showed a recirculation zone at the place where the blockage was being experienced. We had never previously considered recirculation to be a factor.

We learned that particles travelling into this recirculating area spent an inordinate amount of time there until they either stuck to the wall or left the zone due to rapid fluctuations in the turbulent gas flow. Consequently, it is not surprising that solids built up there.

We then had to find some change in the geometry or conditions that would stop the deposition. Although we considered a number of changes, all of them revealed that eliminating the recirculation entirely would have required changing the geometry of the pipe at a very high cost.

The solution turned out to be much simpler. The problem area was located after a section where the flow had been divided into two branches. We found that reducing the flow velocity in the branch with the problem reduced the size of the recirculating area, which in turn reduced the tendency for solids to accumulate. The flow could be increased in the other leg by a corresponding amount without changing the overall flow rate. This cost far less than redesigning the problem geometry, because it involved simply adding an upstream constriction in an area where the pipe had a less complex shape.

Again, we had problems with scepticism within the company, but the ability of the program to pinpoint the problem area helped to convince them. The post-processing results provided by the analysis were a major factor in plant management's decision to install the recommended change at the next process shutdown.

The changes required to reduce the recirculating area were finally made - and they solved the deposition problem. Dramatic time and cost savings were achieved by eliminating the need to shut down the process.

CFD has become a preferred method for analysing deposition problems at Texaco. This technique has greatly reduced the amount of physical testing required to validate proposed fixes and has given us much greater confidence in the solutions that we recommend.

Mr Jung is a research associate at Texaco's R&D department in Beacon, New York

FOR MORE INFORMATION ON: Fluent enter 320