<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[Simon Lab Notes]]></title><description><![CDATA[Notes from Simon Lab on chemical and process engineering design and on treating plant projects as versioned data.]]></description><link>https://simonlab.hashnode.dev</link><generator>RSS for Node</generator><lastBuildDate>Wed, 18 Nov 2026 13:49:55 GMT</lastBuildDate><atom:link href="https://simonlab.hashnode.dev/rss.xml" rel="self" type="application/rss+xml"/><language><![CDATA[en]]></language><ttl>60</ttl><item><title><![CDATA[One tag, three documents, and the list that should stay in sync]]></title><description><![CDATA[Why a plant tag should be an identifier, not a string copied between files]]></description><link>https://simonlab.hashnode.dev/one-tag-three-documents-and-the-list-that-should-stay-in-sync</link><guid isPermaLink="true">https://simonlab.hashnode.dev/one-tag-three-documents-and-the-list-that-should-stay-in-sync</guid><category><![CDATA[engineering]]></category><category><![CDATA[documentation]]></category><category><![CDATA[data]]></category><category><![CDATA[process]]></category><dc:creator><![CDATA[Szymon Wojciechowski]]></dc:creator><pubDate>Tue, 17 Nov 2026 21:36:36 GMT</pubDate><enclosure url="https://cdn.hashnode.com/uploads/covers/6afcc795f0a0b5a97831a56a/7c6d7d9b-2e35-4df7-9ac7-123aa1c32237.jpg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>In a process plant, a tag is supposed to be a stable name for a real object: a vessel, a line, a valve, an instrument, a motor. In practice the tag is often a string that gets copied into many files. The P&amp;ID has it, the equipment list has it, a data sheet has it, an I/O list has it, and an isometric has it again. When those files are edited separately, the tag stops being an identifier and becomes a rumour.</p>
<p>This article looks at the relation between an apparatus, its tag and the documents that ought to mention it. The point is not typography. It is whether a change to one object updates every view that is supposed to describe that object, and whether a reviewer can see the documents that still disagree. The example is synthetic and is taken from public demonstration material, not from a client project.</p>
<h2>A tag is a key, not a label you restyle later</h2>
<p>A useful tag is unique in its class, stable over the life of the project, and used in every document that refers to that object. The vessel is not “the tank on sheet 3” in one place and “T-01” in another, unless those two names are explicitly linked. The line is not “the discharge to the pump” in a calculation and “L-104” on the isometric, unless the calculation states that it read L-104.</p>
<p>Once you treat the tag as a key, a document becomes a view over a set of keys. A P&amp;ID sheet shows some of the objects, with their connections. A line register is a table of the same keys, with medium, size, material and class. An isometric is a fabrication view of one key, with a parts list. An instrument index is another table of keys, each pointing to the line or equipment it belongs to.</p>
<p>The document list for an object is then a derived fact. If you ask “what do we issue for T-101?”, the answer should come from the relationships, not from a memory of which PDF was emailed last week.</p>
<h2>Where the relation breaks</h2>
<p>The break is almost always a hand-copied field. A typical path looks like this:</p>
<ul>
<li>A process engineer adds a nozzle to a vessel on the P&amp;ID and gives an instrument a tag.</li>
<li>An equipment list is updated later, perhaps from an older export, and the new nozzle is missing.</li>
<li>A data sheet still has the previous nozzle schedule.</li>
<li>An I/O list, prepared for the cabinet, never received the new instrument tag.</li>
<li>The 3D model gained the nozzle, but the isometric was issued from a previous extract.</li>
</ul>
<p>Each file can be internally tidy. The project is not. The vessel, the tag and the document list have drifted apart. A reviewer who only opens the latest P&amp;ID will not see that the I/O list is short, and a programmer who only opens the I/O list will not see that the instrument exists.</p>
<p>This is the same “broken chain” problem described more fully here: <a href="https://simon-lab.org/knowledge/project-as-code/">https://simon-lab.org/knowledge/project-as-code/</a>. The chain is not only from discipline to discipline. It is also from an identifier to the set of documents that must stay consistent with it.</p>
<h2>A worked example, kept small</h2>
<p>Consider a synthetic discharge line, called L-104 in public illustrations of this method. Its definition, stated once, includes a line number, a medium, a nominal size, a material, a pipe class, and the equipment it connects. From that definition you should be able to list the documents that read it:</p>
<ul>
<li>the P&amp;ID sheet that shows L-104;</li>
<li>the line register row for L-104;</li>
<li>the isometric of L-104, with its parts list;</li>
<li>any pump or instrument data sheet that takes its process conditions from L-104;</li>
<li>any structural calculation that takes the filled weight of L-104 as a load.</li>
</ul>
<p>Now change one field: the nominal size, from DN50 to DN65. If the tag is a real key, the document list does not change, but every document on that list must be reissued from the new definition. The follow-on checks belong to other keys that depend on L-104. A flowmeter on the line may fall out of range. A pump may remain acceptable, or it may now be oversized. A support may need a new load case.</p>
<p>If instead you only edit the P&amp;ID, the tag L-104 still appears everywhere, and that is exactly what misleads the review. The string matches. The data does not.</p>
<p>The same pattern applies to a vessel tag or an instrument tag. Add a temperature element to the demonstration tank, and the document list should gain or update the instrument index, the hook-up, the I/O list and the cabinet documentation, not only a bubble on the P&amp;ID.</p>
<h2>A document list you can actually review</h2>
<p>A practical review does not start with a folder of PDFs. It starts with an object and two questions:</p>
<ol>
<li>What is the current definition of this tag?</li>
<li>Which issued documents are views of this tag, and which of them were generated from the current definition?</li>
</ol>
<p>A simple table is enough to make the gaps visible. For a synthetic review of L-104 you might record:</p>
<table>
<thead>
<tr>
<th>Document</th>
<th>Number</th>
<th>Issued from revision</th>
<th>Status after the DN change</th>
</tr>
</thead>
<tbody><tr>
<td>P&amp;ID sheet</td>
<td>PID-001</td>
<td>2</td>
<td>Regenerated</td>
</tr>
<tr>
<td>Line register</td>
<td>REG-L-001</td>
<td>2</td>
<td>Regenerated</td>
</tr>
<tr>
<td>Isometric</td>
<td>ISO-L-104</td>
<td>2</td>
<td>Regenerated</td>
</tr>
<tr>
<td>Flowmeter data sheet</td>
<td>DS-FE-104</td>
<td>1</td>
<td>Stale, selection to be repeated</td>
</tr>
<tr>
<td>Support calculation</td>
<td>CAL-ST-012</td>
<td>1</td>
<td>Stale, load to be recalculated</td>
</tr>
</tbody></table>
<p>The value of the table is the stale rows. They are the relation between tag and document list made explicit. In a data-centered project those rows can be produced with the revision. In a file-centered project someone has to build the table by hand, which is already better than assuming that a matching tag means a matching definition.</p>
<h2>What should not be automated quietly</h2>
<p>Generating documents from a single definition removes retyping. It does not choose the new flowmeter, and it does not decide whether the cheaper pump is acceptable to the client. Those are design decisions. The system, or the register, should only be allowed to say that the previous selection no longer satisfies a stated limit, or that a document is stale.</p>
<p>It is also a mistake to treat a tag rename as a cosmetic find-and-replace across files. A rename is a change of key. Historical documents, spare-part lists and operating procedures may still refer to the old tag. The relation has to record the alias, or you have created a new kind of inconsistency.</p>
<h2>Limitations</h2>
<p>The line identifier, the document numbers and the table above are synthetic review material. They are not an export from a production system and not a published file format. The underlying demonstration project, an anonymised NaOH batch dissolution unit, is public example material, not a client reference.</p>
<p>This article does not report error rates, savings or schedule gains from a completed project. Simon Lab’s software platform, which is built around structured project data and generated documents, is in private beta with trusted partners. It is not offered here as a public product, API or self-serve service. A more complete description of the method is at <a href="https://simon-lab.org/knowledge/project-as-code/">https://simon-lab.org/knowledge/project-as-code/</a>.</p>
<hr />
<p><em>This article was drafted with the help of AI and checked by the author, a process engineer with 15 years of design experience.</em></p>
]]></content:encoded></item></channel></rss>