Heat Exchanger Failure Investigation Report

What actually happens when a heat exchanger goes down and you need to figure out why

I spend most of my time walking job sites where a shell-and-tube unit just failed unexpectedly. The operations team wants blame assigned within hours. The plant wants a report before the maintenance window closes. This is the process I use to produce a Heat Exchanger Failure Investigation Report that actually holds up under scrutiny, not just looks pretty for compliance. The first thing you need to understand is that failure investigation is not a paperwork exercise. It is a forensic sequence. You collect evidence before anything gets cleaned, replaced, or altered. I have seen technicians open the tube sheet side and accidentally destroy the very pattern that would have told you how the failure propagated. That mistake is expensive.

Heat Exchanger Failure Investigation Report

Here is the practical workflow I follow. You start with the operating history before the event. Pull the control system logs. Look at temperature differentials, pressure drops, flow rates, and any alarms. Most failures leave a thermal or hydraulic signature weeks before they become catastrophic. If you do not have DCS data, you are working blind. I had one case where the trending data showed a gradual rise in shell-side pressure drop over three months, pointing to progressive fouling on the tube side. The plant thought it was a new batch of feed. It was not. It was a degraded demineralizer upstream that started letting solids through. Next you document the external condition. Look at supports, vibration damage, insulation around nozzles, and any signs of external leakage. Then you open the exchanger in a controlled sequence. Photograph everything before cleaning begins. Seal any openings to prevent contamination. Label heads, channels, and tube bundles so you know exactly where each piece came from. When you inspect the bundle, you need a systematic approach to the tubes. Start with visual examination of the tube sheets on both sides. Look for erosion patterns, deposit buildup, and tube protrusion inconsistencies. Then move to non-destructive testing. I use electromagnetic array testing for rapid screening across the full tube length, followed by hydraulic expansion testing on suspect tubes. For critical service where hydrogen attack or high-temperature corrosion is possible, I take cross-section samples from representative tubes and send them for metallography. That gives you grain structure, carbide precipitation, and crack initiation details that surface inspection alone cannot provide.

Once you have the inspection data, you correlate it with the operating logs. The correlation step is where most reports fail. People list findings without connecting them to root cause. Your investigation needs to show the causal chain. Fouling led to localized overheating, which caused thermal stress, which initiated cracking at the tubesheet roll joints. For the actual report, I structure it with sections on operating conditions prior to failure, inspection methodology, findings by component, metallurgical analysis results, root cause determination, and recommended corrective actions. The root cause section is the part that matters. I state the primary cause clearly, then list any contributing factors separately. Secondary factors should not be dressed up as root causes. They are secondary. If vibration contributed to fatigue cracking, say so, but do not imply it was the main driver just to make the report look thorough. There is a counter-intuitive point that beginners miss. Tube leakage is not always a tube problem. I investigated a case where tubes were leaking and everyone blamed poor water quality. The tubes looked fine. The real issue was that the division plate had shifted, causing the tube side flow to impinge directly on a group of tubes at an angle the design never intended. The impingement erosion created thin spots that eventually leaked. You can miss that if you only look at the tubes themselves.

Another thing worth noting: thermal cleaning with hydroblasting or chemical acid wash can mask failure mechanisms. Acid cleaning removes scale deposits that tell you about the fluid chemistry over time. If you clean before photographing and sampling deposits, you lose the evidence. I once saw a report conclude "normal fouling" on a unit that was actually suffering from microbially influenced corrosion. The biofilm had been washed away during a routine clean, and the investigator never saw the characteristic pitting under the deposits. The corrosion pits only became visible after the deposits were removed carefully in small sections, not by full immersion acid washing. When writing recommendations, be specific. "Improve maintenance" is useless. "Install differential pressure indicators across the tube side and set alarms at 15 percent above normal operating drop" is actionable. Cost estimates help too. A thermocouple upgrade on the outlet header costs roughly two thousand dollars including installation. A full tube bundle replacement with alloy 825 tubes for a unit of that size runs about forty-five thousand dollars parts and labor. People need those numbers to make decisions. The biggest limitation of this whole process is that you rarely get a perfect picture. Tubes that failed may have been removed during prior maintenance. Deposits get lost in cleaning. Operators forget to record certain parameters. I work around this by documenting what is missing and stating how that gaps limits confidence in the conclusion. A report that acknowledges uncertainty is more credible than one that pretends certainty exists. It also protects you when someone later claims you missed something.

If your exchanger is a plate type instead of shell-and-tube, the investigation approach changes significantly. Plate exchangers fail differently. Gasket degradation, plate cracking from overpressure, and channel corrosion are the common modes. The same forensic principle applies, but the inspection methods and root cause pathways are different. I treat gasket materials separately from metal components because the failure analysis for a nitrile gasket hardening at elevated temperature follows a completely different logic than stress corrosion cracking in a 316 stainless plate. I also want to flag a practical bottleneck. NDT availability. Electromagnetic testing equipment is not always on site, and third-party inspection companies have lead times of two to four weeks. If you are investigating a failure that took down a critical process unit, that delay is a problem. In those situations, I prioritize visual and dimensional inspection first, then schedule the NDT while interim operational controls are put in place. You do not need a full report to implement temporary restrictions like reducing operating temperature by fifteen degrees or lowering flow velocity. The final piece is documentation quality. Every photo needs a date stamp and a scale reference. Every sample needs a unique identifier linked to its location on the bundle. Chain of custody matters if the report is ever used in a legal dispute or insurance claim. I have lost track of how many times a report was weakened because a sample could not be definitively placed back on its original tube. It sounds minor until someone asks whether the corroded section you tested came from the inlet zone or the outlet zone.

If you need a template, most engineering firms maintain their own versions. The ones I use follow a structure that mirrors the process I described above. The key is adapting it to the specific failure mode rather than filling in boxes without thinking about the evidence. A report is only as good as the investigation behind it. Garbage in, garbage out applies here the way it applies everywhere else in this work.