Are Fractionator Failings Fixable?

This Month’s Puzzler

We replaced corroded internals and did numerous repairs called for after an inspection of our distillation column. Among the projects, we swapped out pipes and cleaned the feed/product heat exchanger and the reboiler, and put in new control valves and a new feed flowmeter. We’ve now commissioned the unit but are unhappy with the results: product quality from the condensers is marginal; the temperature profile is “unusual” and we can’t increase rate through the tower as expected. I suspect the commissioning work either didn’t follow company guidelines or was poorly managed.

My sleuthing uncovered several issues: 1) pressure readings appear to be lower than expected; 2) the inspector found a number of trays disrupted; 3) that person also discovered several additional broken valves, which apparently has been an on-going problem for the last three years; 4) an inspection by the company responsible for relief valve maintenance detected severe fouling on the relief valve located on the top head — previously, we had cleaned the valves before sending them to the shop, so the inspector didn’t see their as-removed condition; 5) the temperature readings also seem out of whack; 6) the corporate engineer who was responsible reckons the new feed flowmeter is the culprit for the poor tower performance and claims he left its calibration to us. When I asked about hydro-testing and cleaning prior to turning the unit over for commissioning, the supervisor didn’t have any paper work to share; he sent me to the constructors, who said I’d have to get approval from the corporate project engineer. I can’t find much in the files. I talked to maintenance, hoping to find some commissioning reports, but only found a few case files of what they did in the past when they were in on commissioning.

What do you think went wrong? What can be done to get us back on track?

Consider Three Possible Answers

The Puzzler presents many different problems but not really enough details except for some generalities. This reply only addresses three of the six questions posed by the engineer.

Question 1: Why do pressure readings appear to be lower than expected? Column pressure is a measure of vapor mass inventory in the column. Vapor mass inventory balances by reaching a steady state based on enthalpy in (feeds plus reboiler duty) equaling enthalpy out (products plus condenser duty). Column pressure controls adjust some combination of enthalpy flows so pressure is stable at a steady state. If the overhead exchanger is cleaner, a lower differential temperature (DT) is needed to remove the required condenser duty. Some column control configurations will allow the column pressure to drop to reduce the overhead DT and to move the overhead condenser duty to meet the enthalpy balance. If not pressure controlled but allowed to float until enthalpy balance is reached, a lower-than-expected tower pressure is the result of (a) higher condenser capacity than before, (b) lower reboiler capacity than before, and (c) colder feed. The comments here focus on the condenser because of the notation that the overhead relief valves were fouled. Perhaps the overhead exchangers were fouled as well and now are clean?

Question 2: Why did the inspector find disrupted trays? Damaged or blown-out trays never are good for performance. Is there any evidence that trays are disrupted in the current unit? Tower pressure profiles would be one way to catch if trays are blown out. A normal pressure profile for a sieve or valve tray on a 24-in. spacing would show a pressure drop of 0.06–0.15 psi per tray. Numbers might be higher or lower than this if specialized designs are used or in unusual systems. Other ways to verify if trays are disrupted are to look at thermal scans of the vessel or to get a service to gamma scan the column. The three leading causes of tray disruption are flash vaporization due to water entering a hot tower, high liquid levels at startup, and improper installation.

Question 6: Would feed flowmeter miscalibration cause tower problems? This might happen if the calibration was terrible and an attempt was made to run the tower at absolute maximum capacity. However, the flowmeter by itself causing the problem is unlikely. Most conventional trays will run well at rates between 50–100% of design and routinely can be pushed to around 110% of design if standard design margins were used. If the tower has been modified in the past, the original design “allowance” already may have been used and flexibility might be less. Within plant constraints, vary the feed rate by -20% to +10% of normal rates. If performance does not change over this range, it’s unlikely to be a flowmeter problem. At the limit, you can check flowmeter rate by looking at the pressure drop and valve position on feed control valves. This typically will allow you to estimate a rate within 5%.
Andrew Sloley, principal consultant
Advisian, Houston

Evaluate The New Instruments

Consider the following related issues:

1. Causes for not being able to establish normal feed rates to the column could be external to the column, column internals, or both. Conduct a pressure survey to identify sources of flow restrictions. Use one accurate pressure gauge or pressure transmitter and take head pressure corrections into account. Although it would be relatively easier to correct flow problems in the feed line, similar correction in the column could be cost-intensive; locating restrictions in the column interior may require an x-ray of the column. Of course, this could be pricey but will help you spot internal restrictions or errors in installation and subsequent planning for opening the column and repairs.

As an extension to the above, if the feed control valve stays more than 75–80% open at your current flows, the control valve probably is too small. For a short time only, try opening the bypass around the control valve to see if that helps improve flows.

2. Because pressure and temperatures are lower than normal, there could be several external (to the column) causes: flow of heating medium to the reboiler — if the heating medium is steam, you must make sure steam traps are working properly (e.g., for a short time, you may bypass a steam trap to see if steam flow can be increased); a pressure survey around the reboiler could show if there is restriction in the reboiler.

3. The flow meter malfunction could stem from many causes, including, for example, flow meter selection, installation and calibration. Without information on the process liquid, it is difficult to come up with specific recommendations. As a rule of thumb, though, the flow meter should follow upstream and downstream straight run requirements; for “fouling service,” avoid flow meters such as turbine meters and orifice plate/dP cells with impulse lines. Calibration could reveal problems such as zero-offset and non-linearity.

You also hinted at a management issues that need to be corrected. Consider
updating procedures for project execution and responsibilities. Ensure proper records are developed. It may seem, at times, that the documentation slows you down but, in the long-run, proper documentation is priceless and, in some cases, is a regulatory requirement.

Before Operations accepts a project, make sure all inspections have been performed and verified. In one incident I encountered, the cause of premature flooding in a column was traced to inadequate clearance of chimney trays. In the hurry to get the column up and running, this detailed inspection inadvertently had been left out.
GC Shah, senior advisor
Wood, Houston

Focus On Fouling

Clearly some of the problems with the tower have been going on for a while but recent improvements in inspection procedures now are identifying them. Other issues may have been caused by attempts to increase tower throughput. In addition, commissioning errors may have created a fog so that these two categories are difficult to discern.

Based on the fouling at the top of the column, it seems obvious that condenser fouling limits the tower. I’m not sure what fouling has to do with the tray valves but it’s clear that this should be the real goal of the efficiency effort.

Your first step is to capture a sample of the material that fouled the relief valve. After that, sample other areas — focus mostly on the condenser side but don’t leave out the reboiler and feed preheater. Also, sample the trays as well as upstream and downstream streams. If you can’t sample everything now, get what you can and obtain better samples next time; this issue isn’t going away — it’s been there for years.

As for the flow meter, I wonder if it is over-range. I have seen this problem before — where a meter was operating near the maximum and, with an equipment increase, a larger instrument is installed that won’t cover the full range of the measurement unless the throughput increases. The same setback can occur with control valves where they operate poorly at the lower range of their curves.

Obviously, you’re in a political pickle. The corporate engineer blew it and has some control over the plant managers. If you can solve this problem without stepping on any toes, you will win the gratitude of the managers and perhaps even those in corporate. The trick will be avoiding bringing up that farce of a hydro-test and focusing on the fouling problem in the tower. Your next step is to review the procedures and contracts to ensure this type of mistake is prevented.
Dirk Willard, consultant
Wooster, Ohio

November’s Puzzler

We shoot sulfuric acid into our yeast fermenters as a nutrient. When we bought the facility three years ago, it had two fermenters. Since then, we’ve added another fermenter as well as expanded our packaging lines and other parts of the plant (see Figure 1). We now are seeing delays in the injection time that are affecting our yields. We must inject the right amount of acid during a precise ¼-hr window to maximize yield and product quality.