How Traffic Control Plans Actually Get Drawn

There is a wide gap between "we need a traffic control plan" and knowing how one gets produced, and it costs companies real money in two directions — paying an engineer for something a typical application would have covered, and submitting a marked-up aerial for a job that needed a sealed drawing.

Here are the four routes, honestly compared.

Route one: reference a typical application

The traffic control manual publishes standard illustrated layouts for common situations. For a large share of routine work — a shoulder closure, a short-duration lane closure on a two-lane road — setup is accomplished by referencing the applicable standard layout and adjusting it to the site.

This is the fastest and cheapest route, and it is entirely legitimate when the situation genuinely matches the standard case and the agency accepts it.

Two cautions.

Use the currently adopted manual. Part 6 was restructured in the 11th Edition — the current national version being the 11th Edition with Revision 1, dated December 2025 — and states adopt on their own timelines. Under 23 CFR 655.603, states "shall adopt changes issued by the FHWA to the National MUTCD within two years from the effective date of the final rule," and a state manual must conform as a minimum to the national Standard statements. So the document you reference is the adopted one for that jurisdiction, and a reference by old section number may point at nothing. We cover this in the 11th Edition post.

Do not reproduce the figures. Reference them by identifier and link the official manual. The national text is published free at Part 6 of the 11th Edition; state manuals are separately copyrighted and reproducing their diagrams is a different question entirely.

Where this route breaks down: any job where the site geometry, the duration, or the exposure differs materially from the standard case — which is most jobs on high-speed roadways, most jobs with pedestrian complications, and anything an agency has told you needs a drawing.

Route two: a marked-up aerial

The intermediate step. An aerial photograph of the site with devices, signs, and the work space marked on it, plus a note referencing the applicable standard layout.

This is genuinely useful and widely accepted for moderate-complexity work, because it does something a typical application cannot: it shows the actual driveways, intersections, and obstructions the crew will encounter.

It is also the format most likely to draw reviewer comments, because it is easy to produce quickly and easy to produce incompletely. The comments cluster predictably — missing pedestrian routing, no advance warning distances shown, no dimension on the taper, no indication of what happens at the driveway in the middle of the closure. Our post on common reasons plans get rejected covers the pattern.

Route three: CAD or dedicated TCP software

For companies producing plans regularly, a drawing tool pays for itself, and the two options have different characters.

General CAD is flexible and requires someone who knows it. It produces exactly what you draw, including your mistakes, and it has no knowledge of traffic control conventions.

Dedicated TCP software carries device libraries, standard layouts, and often taper calculations. It is faster for this specific job and less flexible for anything else. The evaluation question is whether it reflects the manual edition and state supplement you actually work to — a tool whose library is built on superseded standards produces confident-looking drawings that are wrong in a way that is hard to spot.

The real argument for either is reuse. A company that produces forty similar plans a year and starts each from scratch is spending money on something that should be a template plus site-specific edits. That is the same argument as migrating off spreadsheets in the operational domain: the value is not the tool, it is that the last version of the work is available to the next one.

Route four: an outside engineer

Some jobs require a plan prepared by, or under the supervision of, a licensed engineer, and some require a professional seal. When and where varies by jurisdiction and project, and the question is covered more fully in who can prepare a traffic control plan.

What is not ambiguous is the federal position on the engineering decisions that sit underneath work zone safety on Federal-aid projects. 23 CFR 630.1106 specifies that "an engineer, or an individual working under the supervision of an engineer shall perform" the engineering studies on which work zone safety management strategies are based, and requires those studies to be documented.

That is a decision about agency-level safety strategy rather than about who draws every lane-closure sketch, but it establishes the principle: the judgments the plan encodes are engineering judgments.

Practical advice on using outside engineers: get them the site information they need up front — survey or accurate aerial, actual approach speeds, known obstructions, the agency's specific submission requirements — because the expensive part of an engineered plan is usually the back-and-forth caused by incomplete inputs, not the drawing itself.

Choosing between them

Four questions, in order.

What does the agency require? This settles it more often than anything else. Ask before you decide, not after a rejection.

Does the situation match a standard layout? If the site has an unusual geometry, a pedestrian complication, or a duration that changes the exposure category, the answer is probably no even if it looks close.

What is the exposure? High approach speeds, heavy truck volumes, and workers with no escape route are the conditions under which the plan is carrying the most weight. The crash data is unambiguous that heavy vehicles are overrepresented — the Work Zone Safety Information Clearinghouse records 233 truck-involved fatal work zone crashes in 2024, accounting for 282 deaths, against 850 total work zone fatalities in 763 fatal crashes. Higher exposure justifies more rigor in the plan.

How often will you do this again? One-off complexity favors an outside engineer. Repeated complexity favors building the capability in house.

What a plan has to communicate to the crew

Whichever route produced it, the plan only works if the people building the setup can act on it. Three things determine that.

Legibility in the field. A drawing that is readable on a monitor and unreadable on a phone in daylight is not usable where it is needed. Our post on reading a traffic control plan in the field covers the crew's side.

Dimensions, not just arrangement. A crew can place devices; they cannot infer an advance warning distance from an unscaled sketch.

Availability at setup time. The most common practical failure is not a bad plan. It is a good plan sitting in an email nobody on the truck can find at 5 a.m.

That last point is why plan access belongs with the job record rather than in a folder somewhere. If the current approved plan, the permit, and a photo of a correct setup are attached to the job the crew is dispatched to, the field has what it needs. If they are in a shared drive with a naming convention only the office understands, they do not. That is how Traffic OS handles job documents — the pricing page shows what is included at each tier, and you can book a walkthrough to see it on real screens.

Standing caveat, and it is not boilerplate: requirements vary by state, county, and city — including who may prepare a plan, what format is accepted, whether a seal is required, and how long review takes. Nothing here is legal or engineering advice. Verify with the agency having jurisdiction over your specific project before choosing a route.

The cost nobody budgets

Revision cycles. A plan rejected once and resubmitted costs the drawing time twice plus the review turnaround again, and on a permitted job that delay lands directly on the start date.

The cheapest thing available in this whole area is asking the reviewing agency, before you draw anything, what format they want and what they most commonly send back. Most reviewers will tell you, and most companies never ask.