Sep-2016

Automated column start-ups

Automation at start-up, supported by dynamic simulation, is desirable if it can be accomplished with a modest amount of engineering effort

MARTIN SNEESBY
APESS

Article Summary

Most oil and gas processes run continuously for several years between shutdowns. These plants are typically optimised for ‘normal’ operation and fine-tuned to run up against throughput and/or efficiency limits, often utilising multi-variable controllers and sometimes even dynamic optimisers. The technology is advanced and developed on an ongoing basis. In contrast, start-up procedures can remain virtually unchanged throughout the life of the plant and are rarely considered except in the immediate period around a shutdown or turnaround. At that stage, old procedures and operating instructions that consist of a long series of manually executed steps are dusted off and given to operators who are mostly unfamiliar with the process plant in this transitory state. The prevalence of operating incidents associated with start-ups is, therefore, unsurprising. Kister lists start-up issues as the fourth most common type of column malfunction behind plugging and coking (#1), sump issues (#2) and internals damage (#3).¹ Since that report, the Texas City disaster in 2005 highlighted how easily problems can occur when a column is being operated without its usual automatic controls.

Batch and heavily sequenced processes, as found in some other process engineering sectors, are typically operated quite differently. The focus in these cases is commonly on getting the plant to a particular condition and then keeping it there. Shutdowns and start-ups are often more frequent, and it is more likely that these process transitions will contain elements of automation and be supported by coded sequences.

There are opportunities for aspects of both styles of operation to be incorporated into continuous improvement plans within the other sector. For example, the potential benefits of applying automation (either fully or partly) to start-ups for continuously operated plants include:
1. Safer start-up: better definition of the start-up route and controls to keep the start-up on track; automated detection and handling of possible disturbances; support for operators working in unfamiliar territory.
2. Faster start-up: normal operation achieved quicker, resulting in increased production; less chance of an operating incident that might compromise the start-up.
3. Increased flexibility: freeing up of operating resources to concentrate on other activities during a period of intense activity and pressure; however, the potential need for some process control engineering resources to be on hand might partially offset this benefit.

The total value of these benefits will always be difficult to estimate accurately but the potential avoidance of a single incident could easily be sufficient to justify some investment, or at least an exploration of possibilities. The production benefits (for instance, an extra day of production at full rates) are likely to be enough to sweeten the deal but possibly not more. Yokogawa issued a white paper in conjunction with FRI and claimed time savings of 30% as well as improved operational safety and increased margins to safety and design limits.²

Improved safety is always a worthy goal. Engineers are often required to make a professional judgement with regard to the ‘acceptable level of risk’. Of course, zero risk is not possible and every operating plant carries with it some risk, however well managed. Plant start-ups bring this equation into focus because it is undoubtedly safer not to start the plant at all, yet that philosophy is self-defeating. The real questions are:
• How do I ensure that the risks are fully understood?
• Do I need additional layers of protection against risk?
• If required, what additional layers of protection are practicably possible?
• Do the necessary process operations meet international standards of risk management after all layers of protection are considered?

Costs always need to be considered whenever benefits are being discussed. The costs of a start-up automation project depend on the scope and the execution model. Several approaches are possible. ‘Getting a few heads together’ is always going to be a good starting point, but will it be sufficient given a probable lack of collective experience of infrequent events? And does this method provide enough detail for a control engineer to create an automated sequence? One exciting alternative is to use dynamic simulation or an existing operator training simulator (OTS) to explore and optimise the start-up procedure. The technology to support this approach exists and the right skills are available from specialist and niche suppliers to deliver such a project.

A relatively simple example is considered here: a C4-C5 splitter with non-equilibrium condenser and some thermal integration via preheating of the feed (see Figure 1 and Table 1). However, the principles are extensible and more complex sequences can be developed from the building blocks and methodology. Indeed, a new project is probably best trialled in a test bed arrangement on a relatively small section of plant in order to gain confidence and experience before rolling out across multiple units.

A detailed dynamic simulation model was created for the purposes of studying the start-up in detail. For an operating plant, it would be necessary to validate and tune the model against historical plant data to ensure that the model has sufficient detail to be a good (but not necessarily perfect) representation of the plant and its process dynamics. After validating the model, the simulation must then be migrated to a ‘cold and empty’ condition. This is achieved by performing a process shutdown on the simulation, including venting, draining and purging, where appropriate. The migrated simulation provides a virtual test bed to develop an automated start-up sequence.

The recommended approach from here is to draft a sequence based on typical operating practices with a focus on automation and standardisation. This sequence can then be programmed into the simulation and iterated until an acceptable procedure is established within the simulation environment. The emphasis should be on robustness rather than accuracy.

Ruiz et al described the general principles of distillation column start-up:³
1. The discontinuous phase. This is the initial phase of the start-up, where liquid and vapour inventory is established. The trays are weeping and vapour can bypass the liquid via downcomers. Heating is applied.
2. The semi-continuous phase. Liquid levels are established on the trays and vapour-liquid mass transfer starts. Pressures and levels come under control. Heating continues and the column approaches total reflux operation – stable operation without product flows.
3. The continuous phase. The column reaches the vicinity of the desired steady state. Temperature and composition profiles are established.

Steady state controls become active to drive the column towards its expected operating condition.

A preliminary, generic start-up procedure for this distillation column can be drafted from these principles:
1. Supply liquid to the column to provide an initial inventory.
2. Apply heat to drive temperature and pressure towards normal operation.
3. Apply cooling to generate liquid inventory in the reflux drum and trays.
4. Stabilise operation with total reflux.
5. Establish a suitable composition profile.
6. Move to continuous operation with fresh feed and product draws.
7. Stabilise operation at full throughput.