Texas has been operating under a new traffic control manual since January 18, 2026. The Texas Transportation Commission adopted the 2025 TMUTCD on November 13, 2025, replacing a manual that had been in place since 2011 and revised only twice in the intervening fourteen years — the longest gap between editions in the manual's history.

Six months in, the transition is still incomplete across a lot of plan sets we see. And the target has already moved again: FHWA published Revision 1 to the 11th Edition of the national MUTCD in March 2026, and TxDOT is developing a corresponding Revision 1 to the TMUTCD. Firms that treated the January deadline as a one-time cleanup are going to find themselves doing it twice.

Why the manual changed at all

The driver was federal. The national MUTCD 11th Edition took effect in January 2024, and federal regulation requires each state to adopt a manual in substantial conformance within two years. That gave Texas until January 18, 2026 — a date TxDOT met, though not without complications. When the proposed TMUTCD was posted for public comment in July 2025, several Texas-specific variations from the national manual had been requested but not yet approved by FHWA. TxDOT published for comment anyway, because the federal clock did not stop for the approval process.

The practical consequence is that some of the Texas variations were resolved after the comment period rather than before it. If your reference copy of the proposed manual dates from mid-2025, it is not the adopted document.

What the transition memo actually permits

The compliance provisions are more forgiving than most people assume, and worth reading carefully rather than guessing at.

Traffic control devices purchased or contracted for purchase before the January 18, 2026 effective date may still be installed if they conform to the 2011 TMUTCD Revision 2. Non-compliant devices already in place on existing facilities are not required to be ripped out — they are brought into compliance as part of replacement, maintenance, or upgrading activities. And where a plan set was developed under the old manual, updating it is not automatic; the District Engineer weighs whether change-ordering devices to the 2025 standard would cause unacceptable delay or cost.

This matters for anyone submitting a package right now on a project that was designed under the 2011 manual. The question is not "does this comply with the 2025 TMUTCD" in the abstract. It is a project-specific determination, and it belongs in a conversation with the district rather than in an assumption.

The known-errors problem

One detail that catches people: FHWA does not permit states to correct Standard statements — the mandatory provisions printed in bold — in the national manual, even where an error has been identified. Texas addressed this by including a list of FHWA-known errors as an appendix to the TMUTCD.

If you are working from a Standard statement that seems internally inconsistent or contradicts an adjacent figure, check the appendix before you design around it. The error may already be documented.

What to do about it

Audit your standard sheet library and your specification references against the adopted 2025 manual, not the proposed version. Confirm which of your active projects fall on which side of the effective date, and get the district's position in writing on the ones that straddle it. And build in the expectation of a Revision 1 update rather than treating this cycle as finished.

For firms doing PS&E review work, this is the kind of change that quietly generates comments for two years. The manual is new enough that reviewers are still calibrating, and old enough that "we've always done it this way" is no longer an answer.

There is a gap that opens on large foundation projects, and most owners do not know it exists until they are already standing in it.

The engineer of record designs the foundation. The contractor builds it. The geotechnical firm that did the original subsurface investigation delivered a report months or years earlier and has typically moved on. When conditions in the field differ from what the design assumed — and on deep foundations in variable ground, they routinely do — someone has to make a call about whether what is actually in the hole matches what the design required.

That someone should be working for the owner. Frequently, nobody is.

Why the EOR often declines the role

This is not negligence, and it is worth understanding rather than resenting. Construction-phase observation of foundation installation carries a distinct and substantial liability profile. It requires personnel on site during every shaft, at hours the contractor sets rather than the engineer. It requires real-time judgment calls that cannot be reviewed at leisure. And it exposes the observing engineer to claims from every direction — the contractor if work is rejected, the owner if it is accepted and later underperforms.

Structural EORs frequently and reasonably scope this out of their agreements. Some project documents anticipate the role and require it to be filled. When the EOR declines and the owner does not notice the gap, the role simply goes unfilled, and the contractor's own quality control becomes the only check on the contractor's own work.

What the role actually covers

An owner's geotechnical representative is not an inspector in the checkbox sense. The function is interpretive, and it requires a licensed geotechnical engineer because the judgments are engineering judgments.

On a drilled shaft program that means confirming the bearing stratum is the stratum the design assumed — not merely that the shaft reached the specified tip elevation. It means evaluating the founding surface condition, cleanout adequacy, and water conditions against what the design contemplated. It means documenting variance when the ground does not cooperate, and having the standing to require a design revisit rather than a field improvisation.

Critically, it means the person making these determinations is paid by the owner and reports to the owner. A contractor's quality control staff can be entirely competent and entirely honest and still be structurally unable to serve this function, because the question at issue is frequently whether the contractor's work should be accepted.

The economics run the owner's way

The instinct is to treat this as an added cost. On foundation-heavy projects it is usually the opposite.

Undetected variance in bearing conditions produces one of two outcomes, both expensive. Either the deficiency surfaces later, when remediation means working around a structure that has already been built on top of it, or it does not surface at all and the owner carries an unquantified risk for the life of the asset. Against those outcomes, the cost of continuous engineering representation during installation is small — and it is spent at the only point in the project where correction is still cheap.

There is a second economic argument that gets less attention. An owner's representative with authority to make real-time determinations eliminates the delay cycle that otherwise attends every field question: contractor raises an issue, EOR is consulted, days pass, crews stand down. Having engineering judgment available at the shaft is faster than routing it through an RFI.

Ask early

The time to determine who holds this role is during design, not during mobilization. If your project documents require an owner's geotechnical representative, confirm in writing who is filling it. If they do not, and the foundation program is substantial, ask why not.

The answer is occasionally that the risk genuinely does not justify the role. More often, nobody has asked the question.

For most of the last seventy years, foundation design in Texas ran on the Texas Cone Penetrometer. The TCP charts were quick, they were cheap, they produced a continuous record, and they had a factor of safety baked in. A generation of Texas geotechnical engineers learned to design drilled shafts by reading blow counts off a chart.

That era is over. TxDOT's Geotechnical Manual — LRFD carries TCP in an appendix, under the heading "Legacy Texas Cone Penetration Evaluation." Not a chapter. An appendix, labeled legacy.

The implication is one a lot of practitioners have not fully absorbed: under LRFD, the strength parameters you measure are the strength parameters that drive the design. There is no longer a chart absorbing your uncertainty on your behalf.

What the old charts were actually doing for you

The original TCP correlations date to the 1950s. They provided allowable skin friction and end bearing, and critically, they folded a factor of safety of roughly three into the allowable values. They also — and this is the part worth sitting with — used essentially one relationship across rock types.

That is an enormous simplification. Austin Chalk, Eagle Ford Shale, Taylor Marl, and the Woodbine do not behave alike, and research over the last several years has made clear that a single correlation is most representative of limestone. For weathered and hard shales, the same blow count supports considerably less capacity than the legacy charts suggest.

The safety factor covered that. It covered a lot of things. When your design method carries a factor of three and lumps every rock in Texas into one curve, mediocre subsurface characterization still produces a foundation that stands up. The method was forgiving because it had to be.

LRFD is not forgiving in the same way. Resistance factors are calibrated to a target reliability index against measured parameters. The conservatism is explicit and bounded rather than implicit and generous. Which means the quality of what you measured now propagates directly into what gets built.

What this changes about the boring program

Under the legacy approach, a TCP record at five-foot intervals and strata changes, plus recovery and RQD, was a defensible rock investigation. Under LRFD, that same program frequently is not enough, because the design needs measured strength — and measured strength comes from core.

Three things follow.

Recovery and RQD stop being descriptive and become load-bearing. They were always supposed to be on the log. Now they qualify every strength value you report. A UCS result from an interval with poor recovery is a measurement of the fragments that survived coring, and the fragments that survive coring are the competent ones. Reporting the strength without the recovery beside it overstates the material.

Sampling location matters more than it used to. Strength gradients through weathered profiles can be steep. A specimen recovered two feet above the founding elevation is not necessarily representative of what the shaft will bear on, and under a design method that consumes your number directly, "close enough" propagates.

Formation identification stops being background. When one chart covered all rock, calling the material "limestone" or "shale" was largely a description. When design parameters are formation-dependent and measured, misidentifying weathered chalk as competent chalk changes the answer.

The reporting obligation

If you are producing geotechnical reports for TxDOT work, the parameters section is now the load path. It is worth being explicit about several things that used to be implicit.

State what was measured versus what was correlated, and say which. If TCP data exists — and on many sites it will, either from prior investigations or because it was run alongside coring — reconcile it against the laboratory results rather than presenting both and letting the reader choose. Where they disagree, explain which governs and why.

Report variability, not just central values. LRFD resistance factors are calibrated against an assumed level of characterization. A single value presented without a sense of scatter invites a precision the data may not support.

And be honest about coverage. Borings spaced across a site with known lateral variability characterize the locations they were drilled. Averaging across them produces a number that describes no shaft in particular.

Why this bites hardest on large programs

On a project with a dozen shafts, over-design is cheap and nobody notices. On a program with hundreds, the pressure to sharpen design values is intense, and LRFD gives you the machinery to sharpen them.

That machinery only produces a safe answer if the inputs justify the precision. Tightening a design against thin data is not optimization — it is a transfer of risk from the assumption to the structure, and it does not surface until later.

The legacy charts would have absorbed that. The new method will not. If you are going to design tightly, buy the characterization that earns it: adequate recovery, RQD reported honestly, laboratory strength at the founding elevation, and enough coverage to describe variability rather than assume it away.

The investigation is the cheapest line item on a foundation program and the only one that constrains every other.

Every geotechnical investigation in Texas begins with a legal obligation that a surprising number of project schedules fail to account for. Under Chapter 251 of the Texas Utilities Code, a person intending to excavate must notify a notification center — Texas811 or Lone Star 811 — before breaking ground. Drilling a boring is excavating.

The two-day window is not two days

The statute requires notice no later than 48 hours before excavation begins, excluding Saturdays, Sundays, and legal holidays. In practice that is two working days, and the exclusion is where schedules break.

A ticket submitted Thursday afternoon does not clear Saturday. It clears Monday afternoon at the earliest. Submit before a holiday weekend and the arithmetic gets worse. Crews mobilized on the assumption of a literal 48 hours arrive at a site that is not legally clear to drill, and the cost of a standing crew is the cheapest possible version of that mistake.

There is a ceiling as well as a floor: notice cannot be given earlier than the 14th day before excavation. You cannot clear a site indefinitely in advance and hold the ticket.

What the marks mean, and what they don't

Utility operators mark the approximate location of their facilities. Approximate is the operative word, and it appears in the law rather than in a disclaimer someone added later. Marks indicate where a line is believed to run; they do not certify a precise position, and they do not relieve the excavator of the duty to excavate with care.

There is a second gap that catches people in Texas specifically: water and sewer operators are not required to be members of the notification system. A clear 811 ticket tells you nothing about water and sewer lines. Those operators have to be contacted directly, and on urban sites they are frequently the shallowest facilities present.

Color conventions are standard — red for electric, orange for telecommunications, yellow for gas, green for sewer, blue for water — but the absence of a color does not mean the absence of a facility. It may mean the operator was never notified.

When marks disappear

Locate markings do not survive weather, traffic, or a week of site activity. If marks are destroyed or no longer visible before drilling begins, the correct action is to request a remark rather than to work from memory or photographs.

If a line is struck, nicked, cut, or exposed during drilling, that is a reportable event, and 811 processes it as a high-priority notification. The instinct to quietly assess damage before calling is understandable and wrong.

The exposure

Failure to notify carries civil penalties, but the penalties are not the real risk. Striking an underground facility exposes the excavator to liability for the resulting damages, and on a gas or high-voltage electrical line the resulting damages are not limited to repair cost. There is also the injury exposure to the crew, which no insurance product makes whole.

Build it into the schedule

For firms running frequent drilling programs, the locate requirement is not an occasional formality — it is a recurring gate on every mobilization, and it is one of the few schedule constraints that cannot be compressed by working harder.

The firms that handle it well treat locate submission as a standing item triggered by boring plan approval rather than by mobilization planning. The gap between those two events is usually more than two working days. The gap between mobilization planning and mobilization usually is not.