If you've driven through any DFW suburb and noticed cracked driveways, sticking doors, or foundation gaps, you've seen expansive soil at work. The Eagle Ford Shale and Blackland Prairie formations that underlie much of the Metroplex produce some of the highest plasticity index (PI) soils in Texas — and they move. Seasonally, cyclically, and sometimes dramatically.

At TGG, we investigate sites across Dallas, Tarrant, Collin, and Denton counties on a regular basis. Here's what we've learned from the data.

What Makes DFW Soils So Active

The dominant soil type across much of the Metroplex is CH — fat clay — with Plasticity Index values routinely ranging from 35 to 65. When these soils absorb moisture, they swell. When they dry, they shrink. The vertical movement can exceed 3 to 4 inches over a seasonal cycle in uncontrolled conditions.

The problem isn't just the clay itself — it's the variability. A single residential lot can have CH with PI of 55 in one corner and CL with PI of 28 in another. That differential movement is what cracks foundations, not uniform expansion.

The VOLFRA Method and Why It Matters

TxDOT's pavement design for expansive subgrades uses the VOLFRA (volumetric free swell) method to estimate potential vertical rise. For foundation design on private development, Texas engineering practice typically relies on the PTI (Post-Tensioning Institute) Third Edition methodology, which incorporates the Thornthwaite Moisture Index and edge moisture distance to predict slab behavior.

Both methods are only as good as the soil data that goes into them. A single boring with one Atterberg limit test is not sufficient to characterize a site with variable stratigraphy. We typically recommend a minimum of one boring per 5,000 square feet of foundation footprint, with Atterberg limits, moisture content, and percent passing No. 200 sieve at multiple depths.

Mitigation Options — and Their Tradeoffs

There is no single right answer for expansive soil mitigation. The right solution depends on the structure type, the magnitude of expected movement, the client's risk tolerance, and the construction budget. Common approaches include:

• Drilled pier foundations — extend below the active zone (typically 10 to 15 feet in DFW), isolate the slab from the subgrade. Most reliable for commercial and institutional structures.
• Post-tensioned slab on grade — designed per PTI methodology with edge beams and stiffening beams sized for the site-specific PI and moisture conditions. Standard for residential construction in Texas.
• Lime stabilization — reduces PI by 10 to 20 points through chemical reaction with clay minerals. Cost-effective for pavement subgrades and lightly loaded slabs, but requires adequate curing time and moisture control.
• Moisture conditioning — pre-wetting the subgrade to bring it to near-optimum moisture content before construction. Reduces post-construction swell, but requires time and monitoring.

The Most Common Mistake

The most common mistake we see on residential and light commercial projects is designing for the average PI when the site has significant variability. A PT slab designed for PI = 40 will underperform on a lot where half the subgrade is PI = 60. The geotechnical report needs to characterize the range, not just the mean.

If you're planning a project in the DFW area and want to understand your site's soil conditions before committing to a foundation type, a preliminary site assessment can save significant redesign costs downstream. Contact TGG at info@tggntx.com.

When we tell clients that GeoAI can reduce geotechnical report production time by up to 90%, the first question is usually: "Does that mean you're cutting corners?" The answer is no — and understanding why requires understanding what actually takes time in geotechnical report production.

Where the Time Actually Goes

A typical geotechnical data report for a TxDOT project involves: transcribing boring log data from field sheets into a database, performing USCS classification on each sample, calculating N60 correction factors, generating boring log figures, running design calculations, and then writing a narrative that synthesizes all of it into a coherent technical story. On a project with 10 borings and 200 samples, the transcription and data processing alone can take 20 to 30 hours.

That's what AI addresses. GeoAI's Subsurface Characterization Engine (SCE) extracts TCP and SPT-N values from boring log PDFs, performs USCS auto-classification per ASTM, applies N60 correction per NCHRP 368, detects Shelby tube sample intervals, and generates kriged subsurface cross-sections — automatically. What used to take days takes minutes.

What the AI Actually Does

The report generation in GeoAI uses a structured prompt architecture — not a generic "summarize this" request. The system is built around an 11-section report template that mirrors TxDOT's data report and design report formats. Each section is generated from structured project data, not from the AI's general knowledge. The system includes an iterative QA/QC loop that reviews the narrative for consistency with the underlying data before output.

Critically, no geographic or regional knowledge is hardcoded in the prompts. The AI doesn't "know" that DFW has expansive soils — it knows what your borings found, and it writes about that. The domain knowledge lives in the data, not the model.

What AI Cannot Replace

The licensed PE at the end of the process is not a formality. Geotechnical engineering involves judgment calls that no current AI system can reliably make: recognizing when a boring log anomaly suggests a localized failure plane, deciding whether a thin shale seam is structurally significant, recommending a foundation type that balances performance with constructability in a specific local market. These judgments draw on pattern recognition built over years of project experience in a specific geology.

Every GeoAI output is reviewed and sealed by a licensed Texas PE. The AI accelerates production. The engineer provides accountability.

The Real Value Proposition

The firms that will benefit most from AI in geotechnical practice are not the ones looking to eliminate engineers — it's the ones looking to do more work with the same team. If a senior engineer can spend their time on interpretation and client interaction instead of data transcription, the quality of the deliverable goes up, not down.

That's the case GeoAI is built on. If you're interested in seeing it in action on a real project, reach out at info@tggntx.com.

The PS&E review cycle is one of the most consequential — and frequently delayed — steps in TxDOT project delivery. A package that comes back with major comments at 90% can push a letting date by weeks or months. After analyzing PS&E submissions across multiple project types and districts, the same categories of issues surface repeatedly.

The Most Common Completeness Failures

The required sheet list for a TxDOT highway project varies by project type, complexity, and district. What many EOR firms miss is that the required sheets aren't just defined by the PS&E Preparation Manual — they're also shaped by district-specific preferences, environmental commitments in the environmental document, and utility agreements. A package that is technically complete per the manual may still come back with district comments about missing detail sheets or inadequately addressed utility conflicts.

The most frequent completeness issues we see:

• Missing or incomplete traffic control plans for projects with phased construction
• Summary of Quantities sheet that doesn't reconcile with the bid item list
• Special provisions that reference standard spec items that have been superseded in the current specification year
• Drainage area maps and hydraulic reports not updated to reflect final roadway geometry
• Missing right-of-way parcel maps or parcel descriptions inconsistent with the plans

The Cross-Reference Problem

The most time-consuming review comments are almost never about a single sheet — they're about consistency failures between sheets. A note on a plan sheet that references a detail that doesn't exist. A quantity in the estimate that doesn't match the measurement specified in the plans. A special provision that modifies a pay item that doesn't appear in the bid list.

These cross-reference failures are invisible to a reviewer looking at one sheet at a time. They only surface when someone holds two documents side by side. At high volume, that's impractical. Which is exactly the problem PSEReviewAI's Module 3 was designed to solve.

The Comment Resolution Gap

Perhaps the most frustrating pattern in PS&E review is the partially resolved comment. A reviewer marks a 60% comment as resolved at 90% when the response addresses the specific note but not the underlying issue. This is particularly common in drainage and structural reviews, where a comment about a calculation methodology gets "resolved" by updating the note on the plan sheet without actually revisiting the calculation.

The fix is straightforward in principle: track comments not just as open/closed, but as resolved/verified. At the 90% stage, the review should compare the 90% response directly against the 60% comment text, not just check that a response exists.

What This Means for Your Firm

The firms that move through TxDOT review cycles fastest are the ones with internal QC processes that mirror the external review criteria. Building a pre-submission checklist around the required sheet list, running a cross-reference check between plans and estimate before submittal, and verifying comment resolution systematically — these steps add a day to the internal workflow but eliminate weeks from the review cycle.

PSEReviewAI automates this pre-submission QC. If you'd like to see a demo on a current project, contact info@tggntx.com.

In May 2026, PSEReviewAI became one of a small number of engineering-specific SaaS platforms to receive TX-RAMP Level 2 Provisional Certification from the Texas Department of Information Resources (DIR). If you're a small engineering firm developing software for Texas government clients, here's what you need to know about the program.

What TX-RAMP Is

TX-RAMP (Texas Risk and Authorization Management Program) is Texas's cloud security certification framework, modeled loosely on the federal FedRAMP program. Any cloud service provider that wants to sell to Texas state agencies — including TxDOT, TexDOT divisions, and state universities — is required to obtain TX-RAMP certification before contract award.

There are two certification levels. Level 1 is for low-impact systems (limited sensitive data). Level 2 is for moderate-impact systems — which covers most SaaS platforms that handle agency project data, procurement information, or pre-bid documents. PSEReviewAI handles pre-advertisement PS&E packages including engineer's estimates, which are confidential until bid opening. Level 2 is the appropriate classification.

What the Certification Process Involves

The TX-RAMP process starts with an Assessment and Intake (A&I) questionnaire, which triggers a Provisional Certification if the basic eligibility criteria are met. Provisional Certification is valid for 12 months, during which the provider must complete a Full Assessment covering all applicable NIST 800-53 Rev. 5 controls for the target certification level.

For Level 2, that means demonstrating implementation (or a credible plan for implementation) of 77 moderate baseline controls across 18 control families — access control, incident response, configuration management, personnel security, and more. The full assessment also requires a System Security Plan, Authorization Boundary Diagram, Data Flow Diagram, and Information Security Policy Suite.

What It Means for Small Firms

The honest answer is that TX-RAMP is a significant compliance burden for a small firm. The documentation alone — policies, procedures, diagrams, control narratives — takes weeks to produce properly. And the controls themselves require real infrastructure changes: MFA, formal vulnerability scanning, penetration testing, cybersecurity insurance.

But the barrier is also the point. TX-RAMP creates a credible signal that a small firm has taken security seriously. For a platform like PSEReviewAI that handles agency PS&E data, it's the difference between being considered for agency contracts and being excluded from them. The investment is real. So is the return.

What We Learned

Going through the TX-RAMP process with PSEReviewAI forced us to make infrastructure decisions we might have deferred: proper environment separation, formal access control documentation, a written incident response plan. The platform is more secure for having done it. That's a benefit that accrues to every user of the system, not just Texas state agencies.

If you're a small firm developing engineering SaaS and want to understand the TX-RAMP process, we're happy to share what we learned. Reach out at info@tggntx.com.