He Named My Bridge Pavement Failure Model After Himself — Then the Blowout Inquiry Asked Him to Recalculate the Subgrade Modulus Live

The falling weight deflectometer printout was in the project folder — an A4 sheet from the FWD survey at the bridge-approach section, showing the deflection basin values at 23 measurement stations along the 14km motorway section.

Dr. Abena Mensah-Frimpong had circled three stations in pencil: 7, 12, and 19. Stations where the peak deflection exceeded 487 micrometres — above the 350-micrometre structural intervention threshold — and where the deflection basin shape indicated subgrade weakness rather than surface layer deterioration.

She had written “CBR 3.2% — below design. HDM-4 flag.” in the right margin in her own handwriting.

Marcus was at his workstation, processing the UKPMS rut depth measurements for the current survey cycle — the raw data from the video survey unit that had covered the same 14km section three weeks ago. The UKPMS data was the surface condition record: rut depths, cracking density, texture depth. The FWD data was the structural record: the bearing capacity of the pavement layers, the subgrade condition, the remaining life under traffic loading.

The two datasets were different measures of the same pavement’s deterioration. The UKPMS showed what the surface looked like. The FWD showed what the structure would do under load.

She set the printout on the table between them.

“The three circled stations,” she said, “are where the peak deflection exceeds 487 micrometres. That’s the threshold above which the pavement’s remaining structural life, at current traffic loading, is below eight years. The bearing capacity at those stations is insufficient for the current design traffic level.”

Marcus said: “What’s the design assumption?”

“The original design assumed a subgrade CBR of 5.5%,” she said. “CBR is California Bearing Ratio — a measure of the subgrade’s resistance to deformation under load. The FWD deflection analysis at those three stations implies a subgrade CBR of approximately 3.2%. That’s almost 60% lower than the design assumption. The subgrade here is weaker than the design expected.”

He said: “What does that mean for the maintenance schedule?”

She said: “It means the pavement at the bridge-approach section will reach the 20mm rut depth failure threshold 4 years ahead of the existing maintenance schedule. The existing schedule puts intervention at year 12.

The HDM-4 model, with the corrected subgrade input, puts it at year 8.” She pointed at the circled stations. “These three stations are the failure zone. The bridge-approach section needs to go on the emergency overlay programme.”

She wrote “CBR 3.2% — below design. HDM-4 flag.” in the margin as she said it, pressing slightly harder than usual into the A4 paper. The pencil line was clear.

She put the printout back in the folder. She opened the HDM-4 model on her screen. CIHT-PE-AM-5512 in the project header.

She had submitted the model to Osei two weeks after completing the bridge-approach analysis. He had called her the following day.

He had said: “This is the kind of early-warning analysis that justifies the FWD survey investment.” She had said: “The CBR discrepancy at the bridge-approach is the key — 3.2% versus 5.5% design puts the section 4 years ahead of threshold. That section needs the emergency overlay.” He had said: “This is exactly the intelligence the NH submission needs.”

She had said: “Model is certified under CIHT-PE-AM-5512.” He had said: “Good work, Abena.” She had gone back to the HDM-4 calibration. She had noted: “justifies the investment.” The FWD survey had cost £180,000. She had found what it needed to find. The investment in the survey was justified by what the survey found. What it found was her analysis.

She had not said anything about this at the time. She had gone back to the calibration.

The NH report had been distributed six weeks later. She had read “The Osei Pavement Performance Assessment — pavement condition modelling prepared under Chief Pavement Engineer K. Osei.” She had read “pavement analysis support: Dr. Abena Mensah-Frimpong, CIHT-PE-AM-5512.”

She had opened the project folder. She had taken out the FWD printout. She had looked at her pencil circles at stations 7, 12, and 19. She had read “CBR 3.2%.” She had put the printout back in the folder. She had opened the HDM-4 model output. She had read the 4-year projection. She had closed it.

—

She opened the CIHT registration record. CIHT-PE-AM-5512. She closed it. She took the printout from the folder. She looked at the circles. She looked at “CBR 3.2%.” She put the printout back. She did not annotate it. She opened the next model in her queue.

The CIHT registration — CIHT-PE-AM-5512 — was a Chartered Member designation of the Chartered Institution of Highways and Transportation. The specialist pavement engineering pathway required documented experience in pavement condition monitoring, structural analysis, and maintenance planning, assessed through a competency portfolio and reviewed by a panel of chartered engineers. She had held the registration for nine years.

The HDM-4 software — Highway Development and Management Model, version 4 — was the primary pavement deterioration modelling tool used by highway authorities across the UK and internationally.

It modelled the progression of surface deterioration and structural failure under traffic loading and environmental conditions, using calibrated deterioration equations derived from long-term pavement research. She had used HDM-4 on 23 pavement management assignments.

The model’s accuracy depended on the quality of the input data — the UKPMS surface condition records, the FWD structural data, and the traffic loading spectrum — and on the calibration of the deterioration equations to the specific pavement type and climate of the section being modelled.

The bridge-approach calibration had been the most complex she had done. The deflection basin shape at stations 7, 12, and 19 had shown a pattern she had not seen before — the peak deflections were above the critical threshold, but the basin shape was atypical.

The centre-point deflection was not as high relative to the offset deflections as it would be if the problem were in the surface layers. The basin shape indicated a deeper failure: the subgrade, not the asphalt layers. She had run the back-calculation analysis three times before settling on the CBR estimate of 3.2%.

The back-calculation was a numerical inversion of the deflection basin measurements to estimate the layer moduli of the pavement structure. She had used the ELMOD software for the back-calculation — the standard tool for FWD analysis in the UK highway engineering community. CIHT-PE-AM-5512 in the model output header.

The Highways Maintenance Conference was in Birmingham — two days, 220 delegates, sessions on pavement condition monitoring, whole-life costing, and emergency intervention planning. Osei presented in the third session: “Early-Warning Pavement Deterioration Modelling: A Case Study in Bridge-Approach Structural Analysis.”

He had the HDM-4 rut depth projection curves on his slide 4 — her curves, from her HDM-4 model, showing the 20mm failure threshold crossing at year 8 instead of year 12 under the corrected subgrade CBR input.

He had the bridge-approach risk zone marked on the highway alignment map — her annotated alignment map, the circle she had drawn around the bridge-approach section in the initial model presentation.

He said: “Our pavement assessment methodology identified a 4-year ahead-of-schedule intervention requirement at the bridge-approach section — a critical finding that enabled early emergency overlay planning before structural failure.”

He said “our pavement assessment methodology.”

He did not name HDM-4. He did not name the FWD deflection basin analysis or the subgrade CBR discrepancy that had produced the 4-year finding. He did not name the CIHT registration. He did not name Abena.

She had read the conference citation in the Highways England technical newsletter two weeks after the event. She had noted it. She had gone back to the new section model.

The contact from Dr. Yvonne Hartley had arrived in her professional email.

“Dr. Mensah-Frimpong — I am the Transport Select Committee inquiry’s technical specialist for the Arroyo Seco bridge-approach incident. A tyre blowout at station 12 of the 14km motorway section on [date] has resulted in a police accident investigation and a parliamentary inquiry into the National Highways maintenance decision for that section.

The inquiry requires:  the original pavement condition model for the 14km section, with the CIHT-registered pavement analyst’s certification; (2) documentation of CIHT-PE-AM-5512 registration; and  the analyst’s availability for testimony before the Transport Select Committee. The CIHT registration number on the model documentation identifies your registration. Please confirm your availability.”

She read “CIHT-registered pavement analyst.”

She read “station 12.”

Station 12 was one of her circled stations on the FWD printout. Station 12 was where peak deflection had been 512 micrometres — above the 487-micrometre critical threshold she had identified, above the 350-micrometre structural intervention threshold. She had circled station 12 in pencil on the FWD printout in her project folder.

She opened the CIHT registration record. CIHT-PE-AM-5512. She took the FWD printout from the folder. She looked at the circles at stations 7, 12, and 19. She looked at “CBR 3.2%.” She looked at station 12’s deflection value — 512 micrometres — in her pencil annotation at the side of the circle. She put the printout back in the folder.

She did not call Osei.

She opened a reply to Dr. Hartley. She confirmed her availability. She attached: the HDM-4 model documentation (all 47 saved versions of the project file, with the CIHT-PE-AM-5512 certification header in each), the CIHT registration certificate and scope document, the FWD survey data for the bridge-approach section, and a four-page timeline document she had drafted — the model’s history:

the FWD survey date, the analysis completion date, the NH report submission, the 4-year ahead-of-schedule finding, the bridge-approach emergency overlay recommendation, and the timeline of events from the model to the blowout incident.

The timeline had taken her two hours to write.

She sent the package to Dr. Hartley. She went back to the current model.

The incident notification had arrived at the Authority’s central office and been forwarded to Osei the same day. He had read it. He had read the TSC inquiry notification that had come with it.

He had told the Authority’s legal team to prepare the pavement condition evidence file.

The legal team had come back within the hour. “The Transport Select Committee inquiry has contacted Dr. Mensah-Frimpong directly. CIHT-PE-AM-5512 is her registration. She confirmed her availability this afternoon and sent the model documentation.

The inquiry testimony requires the CIHT-registered pavement analyst — that’s Dr. Mensah-Frimpong. Your CEng is highway engineering management, not CIHT MCIHT pavement fatigue analysis. You cannot be cross-examined on FWD deflection basin methodology or HDM-4 calibration.”

He had looked at the NH report on his screen. “Osei Pavement Performance Assessment.”

He had said nothing for a long time.

The legal team had said: “She sent the documentation this afternoon. She didn’t contact you first.”

He had said: “No.” He had not said anything else.

He had been Chief Pavement Engineer for twelve years. He had managed 34 NH pavement condition submissions — annual reports covering hundreds of kilometres of motorway section, each submission built from the data his pavement analysis team produced. He had submitted every one. He was the Authority’s designated NH technical correspondent. He signed the reports. He submitted them to National Highways.

The pavement analysis team had changed over twelve years. Different analysts at different times, with different specialisms, different CIHT registrations. He had always been the submitting engineer. The NH submission was his report, assembled from the analysis work his team conducted. The team’s work was how he knew the condition of the sections he managed.

He had never thought of the HDM-4 model as Abena’s model specifically. He had thought of it as the Authority’s bridge-approach pavement model — the tool his pavement maintenance programme had deployed to identify the intervention requirement at the section.

The model existed because his programme had funded the FWD survey, specified the analysis, and commissioned the HDM-4 calibration. The model was the analytical output of the programme’s technical framework.

But the inquiry was not asking about the programme’s framework. The inquiry was asking about the model — the specific HDM-4 calibration, the subgrade CBR input correction, the 4-year ahead-of-schedule finding. The TSC technical specialist wanted to understand how the model had predicted the structural failure timing, and whether the prediction had been communicated to the maintenance scheduling team before the blowout occurred.

He could not explain the HDM-4 calibration methodology because he had not conducted it. He knew the calibrated model’s output — the 4-year finding, the emergency overlay recommendation — because the report had told him. He had received the finding. He had used the finding. He had submitted the finding to National Highways as the NH report.

But knowing the output of a model and being able to explain under examination how the model had produced the output were not the same thing. The TSC would not ask him about the emergency overlay recommendation. They would ask why the model had predicted failure at year 8 instead of year 12.

The answer required explaining the FWD deflection basin analysis, the subgrade CBR inference, and the HDM-4 fatigue deterioration equations with the corrected CBR input. That answer was Abena’s.

He had said “justifies the FWD survey investment” when the model had confirmed the early intervention need. He had meant it as recognition of the survey’s value — the £180,000 FWD survey that had produced the data that had produced the finding.

He had connected the investment to the finding. But the model that had made the connection — that had looked at the deflection basin shape at stations 7, 12, and 19 and inferred a subgrade CBR of 3.2% rather than the design assumption of 5.5% — was not the investment. The investment had made the data possible. The model had made the meaning possible.

He opened the NH report amendment form.

The FWD printout was in the project folder on her desk. She had put it back after the TSC contact. She was at her computer, working on the inquiry documentation — confirming the model file inventory, checking the CIHT certification documentation against the inquiry’s requirements, preparing the timeline narrative. The folder was on her desk, to the left of the keyboard. The printout was inside the folder. Her pencil circles were on the printout. The “CBR 3.2%” annotation was in the margin.

She had not taken the printout out again since putting it back. She was working through the documentation items. The printout was in the folder. When she needed it for the inquiry hearing, she would take it out.

She finished the documentation package. She sent it to Dr. Hartley with a covering email: all items as requested, plus the timeline narrative, plus the model validation documentation showing the HDM-4 calibration against the observed rut depth progression in the first three years after the bridge-approach section had been built. The validation showed that the calibrated model had accurately tracked the section’s actual deterioration before the FWD survey had been conducted.

The covering email was eight sentences. She sent it. She went to the current model.

The Transport Select Committee inquiry was held in a committee room at the Palace of Westminster — the standard layout: the Committee members across the central table, witness chair at one end, parliamentary officials at the sides, public gallery behind. Abena was the first technical witness of the session. Osei was scheduled as the second — to speak on the Authority’s maintenance programme management and the NH report submission process.

Osei’s introduction was four sentences. He described the Authority’s pavement maintenance programme, named Abena as the CIHT-registered pavement analyst who had built the bridge-approach model, stated that all technical questions about the model’s methodology were for her, and confirmed he would be available to discuss the NH report’s submission and distribution. He did not speak on technical methodology for the remainder of the session.

The Committee Chair led the examination with the fundamental question: “Dr. Mensah-Frimpong, when did your model predict that the bridge-approach section at station 12 would reach structural failure, and what was the basis for that prediction?”

She had the FWD printout unrolled on the table in front of her. She pointed to station 12.

“The model predicted structural failure at year 8 from the survey date — 4 years ahead of the existing maintenance schedule. The basis was the FWD deflection basin analysis at station 12, which showed a peak deflection of 512 micrometres. That’s above the 487-micrometre threshold I identified as critical for this section.”

She pointed to the circle at station 12. Her pencil line. Her handwriting in the margin.

“The deflection basin shape at station 12 implies a subgrade CBR of approximately 3.2%. The design assumption was 5.5%. A subgrade at 3.2% deteriorates significantly faster than the maintenance schedule was designed for.

I ran the HDM-4 fatigue deterioration model with the corrected subgrade input. The model produced the year 8 failure threshold. I reported this to the Authority’s pavement maintenance programme with an emergency overlay recommendation for the bridge-approach section.”

The technical specialist, Dr. Hartley, asked: “When was the emergency overlay recommendation submitted, and what action was taken?”

She said: “The recommendation was in the model report submitted to the Authority on [date]. The emergency overlay was not scheduled before the blowout incident. I don’t have visibility into the scheduling decision — that’s the Authority’s maintenance programme management.”

She answered 34 questions over three and a half hours. The Committee examined the FWD survey methodology, the HDM-4 calibration process, the CBR inference from deflection basin analysis, the emergency overlay recommendation, the NH report submission, and the timeline from the model to the incident. She answered from the model documentation — the 47 saved HDM-4 project files, the FWD survey data, the timeline narrative she had prepared.

The inquiry record built throughout the session. The Committee Clerk typed continuously. By the time the session concluded:

“CIHT Registered Pavement Analyst: Dr. Abena Mensah-Frimpong, CIHT-PE-AM-5512. HDM-4 model predicted failure at year 8, station 12 peak deflection 512 micrometres. Emergency overlay recommended [date]. Pavement failure threshold reached [date of incident].”

Dr. Hartley emailed her after the session.

“Dr. Mensah-Frimpong — your CIHT registration and your HDM-4 methodology are the technical foundation of this inquiry. The station 12 identification and the 4-year ahead-of-schedule prediction are the finding. The record is clear.”

Marcus had been waiting at the office when she returned. He said: “CIHT-PE-AM-5512 in the parliamentary record.”

She said: “Yes.”

He said: “Station 12.” He said it the way you say the name of a place where something happened.

She said: “Station 12.” She took the FWD printout from the table. She looked at the circle at station 12.

Osei called that evening. He said: “The inquiry outcome is satisfactory. Your model was the technical basis for the Committee’s finding.” She said: “The HDM-4 calibration was complete.” He said: “Yes. I’ve amended the NH report — your name and CIHT registration. And I’m implementing a protocol: CIHT-registered pavement analysts named on all NH pavement condition submissions going forward.” She said: “Yes.” He said: “Good work, Abena.” She said: “Thank you.” She put the FWD printout back in the folder. She opened the next project file.

She did not say anything else. She opened the next project file and began reviewing the new FWD survey data.

The 34 questions had covered the full scope of the inquiry’s technical concerns. The early questions had been foundational — what FWD measures, what HDM-4 does, what CBR means and why it matters for pavement design. These were the questions the lay Committee members needed to understand before the technical findings could be useful to them.

She had answered each one without abbreviating the explanation. The Committee members were not pavement engineers. They were parliamentarians conducting a safety inquiry. They needed to understand the technical content well enough to evaluate the maintenance decision.

The later questions had been more specific. The TSC technical specialist Dr. Hartley had driven the technical examination: the ELMOD back-calculation, the CBR inference methodology, the HDM-4 calibration validation, the basis for the 4-year ahead-of-schedule finding, the documentation of the emergency overlay recommendation.

Each question had required a complete technical answer, drawn from the model documentation and the FWD data. She had answered each from the record.

The Committee Chair had asked the question she had known was coming, at the end of the technical session: “Dr. Mensah-Frimpong, in your professional judgment, if the emergency overlay had been implemented before the incident, what is your assessment of whether the blowout would have occurred?”

She had said: “I can’t make a definitive causal statement. What I can say is that the model predicted structural failure at station 12 within the existing maintenance schedule window. An emergency overlay would have addressed the subgrade weakness by reinforcing the pavement structure above the threshold level. The deflection at station 12 at the time of the incident, if measured then, would have been above the structural intervention threshold I identified in the original analysis.”

She had not said more than that. She had returned to the documentation.

The inquiry record was in the parliamentary archive. It would be permanent. CIHT-PE-AM-5512. Station 12. She had the parliamentary session reference number.

Osei’s statement to the inquiry — his testimony after hers — had addressed the NH report submission process and the maintenance scheduling decision. He had described the Authority’s emergency overlay programme and the competing demands on the maintenance budget in the year the model had been submitted.

He had stated that the bridge-approach section had been on the emergency overlay priority list, but that the budget allocation for that year had been insufficient to cover all priority sections.

She had not been in the room for his testimony. She had left after her session ended.

She had read the published TSC inquiry report three weeks later. Her technical findings were in the report. Osei’s evidence on the scheduling decision was in the report. The conclusion was in the report:

the bridge-approach section should have been scheduled for emergency overlay in the financial year following the model’s submission. It had not been. The inquiry recommended revised NH pavement assessment protocols requiring CIHT-registered pavement analyst named authorship on all submissions.

She had read the recommendation. She had filed the report. She had gone back to the new motorway section analysis.

She was preparing the new FWD survey data — a different motorway section, a different deflection basin pattern, a different calibration baseline. Before loading the new data, she took the bridge-approach FWD printout from the old project folder — the A4 sheet with her pencil circles at stations 7, 12, and 19, the “CBR 3.2% — below design.

HDM-4 flag” annotation in the margin. She used it as a calibration reference: comparing the new section’s peak deflection values against the known threshold-failure station 12 baseline before running the new HDM-4 model. Marcus was at the UKPMS data processing station, extracting the new rut depth measurements. The inquiry record was in the Transport Select Committee archive — “CIHT Registered Pavement Analyst:

Dr. Abena Mensah-Frimpong, CIHT-PE-AM-5512, HDM-4 4-year ahead-of-schedule prediction, station 12 FWD.” She spread the new survey data beside the old printout. She took out the printout. She looked at the circles. She opened the new calibration file.

The amended NH report had arrived by email — “Pavement model by Dr. Abena Mensah-Frimpong, MCIHT, CIHT-PE-AM-5512.” She had read it. She had filed it in the bridge-approach project folder alongside the original NH report with “pavement analysis support.” Both were in the folder.

The Authority protocol had arrived the same day — the formal document requiring CIHT-registered pavement analyst named authorship on all NH pavement condition report submissions. She had read it. She had filed it alongside the amended NH report.

The NH technical bulletin — “Osei’s pavement assessment” — had been distributed to 34 regional engineers before the inquiry. The bulletin was in their files. The inquiry outcome and the amended NH report had been distributed to the same 34 engineers. She did not know which had updated their working files. She had the bulletin reference number: NHS-PAV-2024-BR-12.

The new motorway section’s FWD data was from a 9km section in the north of the region — different geology, different traffic loading, different maintenance history. The deflection basin values were generally lower than the bridge-approach section’s had been.

Only two stations were above the 350-micrometre structural intervention threshold — stations 4 and 7 of the new section — and the peak values, at 378 and 402 micrometres respectively, were below the critical 487-micrometre level she had identified on the bridge-approach section.

She noted the two stations in the project log. She would run the HDM-4 model with the new data and check the remaining structural life, but the preliminary assessment was that this section was within its design life.

Marcus brought the UKPMS rut depth printout. She placed it beside the FWD data — the surface condition record alongside the structural record. The rut depth readings at the two flagged stations were moderate, consistent with the deflection basin values.

She loaded the new data into the HDM-4 model. She set the calibration parameters — the traffic loading spectrum, the environmental inputs, the pavement structure data from the design records. CIHT-PE-AM-5512 in the project header.

She had not thought about the bridge-approach inquiry since the day she had returned from Westminster. She had thought about station 12 the evening of the hearing, when she had been driving home.

She had thought about the 512-micrometre deflection value and the emergency overlay recommendation and the maintenance scheduling decision. Then she had been at her front door. She had not thought about it again. She had been at her desk the following morning, reviewing new UKPMS data, and it had not come into her mind.

She ran the first HDM-4 model iteration. The results came back within three minutes. She reviewed the structural life projections for each station. Stations 4 and 7 had remaining lives of 6.2 years and 5.8 years respectively — below the 8-year trigger for emergency overlay consideration, but above the 5-year threshold for priority scheduling.

She noted the findings. She prepared the preliminary model output for review.

She took out the printout. She looked at the circles. She opened the new calibration file.

The new section’s preliminary report was ready after two HDM-4 iterations. She had confirmed the structural life projections for stations 4 and 7, cross-checked them against the UKPMS surface condition data, and verified that the deflection basin shapes at both stations were consistent with surface layer deterioration rather than subgrade weakness. Different mechanism from the bridge-approach.

The pavement structure at the new section was behaving as the design expected. The two stations that had triggered the deflection threshold were flagged for priority scheduling in the standard maintenance cycle — not emergency overlay, priority scheduling. She noted the distinction in the model report. CIHT-PE-AM-5512 in the report header.

She had written the preliminary model report in 45 minutes. She sent it to Osei’s office for review — the new section was within his maintenance programme jurisdiction. She had not spoken to Osei since his call after the TSC inquiry session. The report was the communication. She had addressed it to the programme and sent it.

Marcus brought the next FWD dataset. A third section — 6km, medium traffic loading, the deflection data from last week’s survey. She took the dataset. She opened the ELMOD software. She began the back-calculation.

She took out the printout. She looked at the circles. She opened the new calibration file.