Plate Heat-Exchangers – Exposing a huge design fault - The Thin Blue Line - and technical solution

The Thin Blue Line

This article exposes a fundamental design flaw in the design and selection of plate heat-exchangers (PHE). Thankfully, we have developed a technical solution to this challenge!

Background

We performed a large data analysis of extremely large data sets in order to check the thermo-fluid performance predicted by the selection software program from one of the world's largest and most prestigious plate heat-exchanger manufacturers, as well as a number of their competitors.

This article discusses the surprising findings we discovered.

We found that the fundamental theory underlying the design, selecting and optimizing of plate heat-exchangers, has a hidden flaw. This flaw then often results in a huge number of off-optimum, or sub-optimum heat-exchangers. These are seen in the sea of dark blue dots, on the plot shown in Figure 1 below.

Actually, it turns out that there are at least two fundamental technology oversights.

The Thin Blue Line

Figure 1 - The Thin Blue Line

Large Data Project

During the large project analysis of the original equipment manufacturer (OEM) and competitors' data we used the design/selection software generally used by their engineers and often supplied to their customers. Results were generally similar for all of their competitors as well.

How could this be the case?

Industry sources informed our team that all manufacturers trace their software back to the same original technology source. Engineers that worked for the previously mentioned large OEM apparently eventually left and allowed this know-how to pass into the rest of the industry over a number of years. So, this line of technical continuity conveyed the flaw/s onwards.

Some of the manufacturers had converted the original algorithms into alternative computer languages and used these as the core technology for their calculation engines. Yet, the underlying technology flaws were not discovered.

To their credit, a big-data study performed from outside the software code, often allows patterns to be observed that the software designers had not envisaged from within. This a property of big-data systems. Fascinating. 

Data Extraction Process

A program was first developed to build reasonable, rational thermo-fluid data sets, incorporating mass flow-rates, terminal temperatures, heat-transfer, allowable pressure drops.  

These data sets were injected into the software and the output captured through a variety of technologies. This data was stored in a database. At some point, we had collected 105,000 data points.

We found that managing this amount of data can be somewhat problematic for a standard spreadsheet. Excel fails at something like 30,000-40,000 data points, but LibreOffice on Linux can manage 60,000 points with ease.

Visualization software was written in order to visualize the data effectively, and to assist with pattern recognition.

Machine-Learning Database

A copy of a stripped-down version of the 60,000 data point database was sent up to a prominent machine-learning data-science research forum, around six years ago, for data scientists to experiment with. Apparently it lay essentially dormant until recently. In the past few weeks, we had a number of enquiries from researchers around the world, in regards to this database, after six years of apparent silence. This was surprising, yet gratifying. Frankly, we had forgotten the data-set was still there.

So, we added in a few more computed columns of data to that data-set, clarified the meaning of each header tag, and uploaded an updated version to see what other additional optimization patterns may emerge.

Further research

We are conducting further research internally on this dataset, in order to uncover further patterns to be used in our optimization work.

Thus far, we have discovered two specific laws that govern designing and selection of an optimized plate heat-exchanger.

Optimization Consulting

Using our specific research, we can determine how close to optimum a particular plate channel solution lies. These laws then add in further filters in guiding the designer towards an optimum solution.

Application of these laws on top of the legacy software optimization rules established many years ago (long in the tooth by now) - allow a vastly refined, optimized result to emerge.

The net effect is to remove most of the blue spots seen in the 'Thin Blue Line' picture, and push them onto optimum lines.

Technology Offer

• Existing plate manufacturers and OEM's
• We can work with your thermo-fluid engineers and software developers to implement these new rules, in order to provide you a massive size and cost advantage over your competitors.
  
  
• PHE designers, engineers and re-builders
• We can provide you a simple technical basis to check whether your software package is actually giving you an optimum solution, or not.
• If not, then you can continue your design selection until you arrive at a near-optimum solution.
  
  
• We offer a performance guarantee on our technology!!!

We look forward to your inquiries.

Best regards,
Desmond Aubery

Managing Partner
Rarefied Technologies (SE Asia) - RareTechSEA

https://sites.google.com/view/raretech-sea/technology/technical-consulting