Roadside Barriers – are they just barriers?

In another gripping Road Ahead exclusive from renowned forensics specialist Stan Bezuidenhout, we get the inside track on the underlying defects of certain roadside barriers – which can be fatal when things go wrong

On 23 August 2008, a lovely young lady – let’s call her Lisa (this is an actual case, but names are changed to protect the parties involved) – was travelling on the M3 near the University of Cape Town on a late Saturday afternoon on her way home.

Lisa was alone in her little car, a white Renault Clio.  There was a Land Rover on the road as well.  It was driving just ahead of her, in the right-head (fast) lane.  Lisa was driving in the left-hand (slow) lane, furthest from the barrier that separated her roadway from traffic heading in the opposite direction.

The earlier rain had stopped, but the roads wet still we in places and the smell of the after-rain air, enriched by Newlands Forest, was discernible.  On that same day, the Lions and the Bull had clashed in a rugby match worth remembering.

Lisa might have had plans for later in the day, aspirations for her future or concerns about her life.  We will never know, as those likely happy thoughts would be obliterated forever.

As the Land Rover driver testified during the trial that followed, a silver Ford Fiesta approached from the opposite side and crashed through the roadside barriers, back side first, and rammed onto the roadway where Lisa and the witness were driving.  The roadway for the Fiesta was only 540mm higher than that where Lisa and the Land Rover were driving, but the car landed on their side of the road and came to a standstill almost immediately.

The Land Rover witness testified that she had believed the Fiesta was going to collide with her. But fortunately, she had been able to move over immediately to the left and to safety.  The witness looked in her rear-view mirror as she was leaving the carnage behind and saw Lisa’s Renault pass seemingly safely by the Fiesta, slow down and stop on the left side of the road a distance away.

The witness said a silent prayer of thanks for escaping the worst and thought nothing further of it.  In her mind, Lisa was simply pulling over and stopping.  Like many others, this witness had carried on driving, believing a disaster had been averted.  However, she had no idea of just how wrong she, and the others, would be.  According to police reports, the Ford Fiesta had been speeding, lost control and crashed through the barrier, heading straight for Lisa and the Land Rover.  Words such as “speeding”,
“very fast”, “racing”, and “flying through the air” were some of the graphic illustrations of the perception of witnesses.  A colonel from the National Office of the South African Police Service came to testify that the Ford Fiesta had been travelling so fast that it had lost control, and that its speed had been so extreme it had had enough energy to crash right through the roadside barrier and onto the road below.

Another witness, driving a black Mercedes-Benz SLK, testified the Fiesta overtook her on the left at high speed, and that the Fiesta then lost control right in front of her as they were going through the slight right-hand bend.  She further testified that the Fiesta then “flew through the air”, and that she saw flames from underneath the car as it did so, with debris from the terrible crash flying onto her car and causing damage.

The prosecution was content:  They could argue that speed and tunnel vision had set in.  They targeted a suspect and they believed he had been speeding.  They were going to prove this – no matter what.

Fortunately, the other car landed on the same roadway where Lisa was travelling.  Her luck was unbelievable, since the Fiesta never collided with her – but it was also short-lived, as the car’s crashing through the barriers caused something like a high rock to shatter her windshield and enter her vehicle, striking her in the head and fatally injuring her.

The assault had been so violent and so sudden that police reported her vehicle had rolled on for almost 400m before coming to a halt on the side of the road.

The first witness on the scene reported that, when they got to her, she was still drawing her last breath.  Witnesses watched as all Lisa’s aspirations, thoughts, dreams, and desires slowly dwindled as her heart slowed down and her breaths became ever shallower.

The paramedics declared her dead at the scene and the postmortem report confirmed she was killed by massive blunt-force trauma to the head.  She was clutched from the caring hands of her family and friends in a way that no one could have predicted or planned for.

According to all witnesses, the state prosecution and the police reports, the driver of the Fiesta had been the cause of her death.  According to the prosecution, with support of witness testimony, the driver had lost control when entering a bend at a high speed and that it was this speed that had caused him to lose control and crash through the barriers.  In so doing, the enormous forces had shot up stones and debris, one piece of which had caused Lisa’s fatal injury.

They had the Fiesta driver dead to rights.  The prosecution was confident from the start.  With a Gerrie Nel-Style of aggression, the prosecutor attacked the case, the driver of the Fiesta’s version as well as the expert witness for the defense:  Stan Bezuidenhout from IBF Investigations.  Those attacks included an assault on his qualification.  “So, what qualification do you actually have?”  He was also challenged on the number of cases he had attended to in his, then, 12-year career.

As Bezuidenhout testified, the cracks began to show in the state’s case and their confidence started to dwindle along with it.  It was when he addressed the issues, he had with the mistakes made in the SAPS investigation, that the case started becoming stranger than fiction.  He was not allowed to continue with his testimony and asked to continue at a later point in his report, some 10 pages further.  The court would hear nothing of errors made by the state.

But this article is not about the defense’s strategies and efforts.  As the trial dragged on, one witness testified he saw the Ford Fiesta and a black Mercedes SLK seemingly racing each other swerving in and out of traffic as they changed lanes, until he lost sight of them some 5 km from the scene of the collision.  He saw them again at an intersection, still one behind the other, and not again until he passed the scene of the collision.  He was left with the impression that the Ford had lost control and crashed through the barrier – another nail in the Fiesta driver’s coffin.

It was when the driver of the Mercedes SLK testified that the first conflicts in evidence began showing.  She claimed she never raced against the Ford, that she never saw him until the collision and that she was driving in the area of the collision when this Ford overtook her at high speed., on her left-hand side, entered the bend, lost control and crashed through the barrier and onto the road on the opposite side.  She even testified she saw the Fiesta “leap through the air, partially ablaze”, before crashing onto the road below.  She specifically testified that “the crash was so severe that debris was thrown backward and onto her vehicle, before she could pull over and stop on the left-hand side of the road.”

While the police investigators had been called to the scene, they failed to arrive in time.  The first police officers arrived after the driver of the Ford Fiesta had been removed by an ambulance; the first photographs of the scene had only been taken after the Ford Fiesta had already been removed by tow-in services, with permission from the first arriving police officers.

Scene at night – poor quality as original

In their statements, police investigators claimed they had taken the photographs between 17h30 and 20h20. According to sun position data sourced by Bezuidenhout, it would still have been light when this had been done – but the police photographs were clearly all taken in complete darkness.  What had the officers then been doing during the daylight portion of their investigations, when they could have produced better images?

6.4    It is not possible to scientifically calculate the speed of any of these vehicles at the time of the incident with the available information.
6.5   These conclusions and endings were made based on the information that were given to this office.  Should more photographs or additional information become available, we reserve the right to evaluate such information in order to give a supplementary statement with regards to the new information.

This was the first clue that something was amiss, and Bezuidenhout immediately noticed.  The chain of evidence had been broken.

As a former specialist reservist in the South African Police, and with his private industry experience, Bezuidenhout know something was wrong with the prosecution’s evidence.


Damage to Lisa’s car – poor as in original police photo


Damage to Lisa’s car – poor as in original police photo

During trial, it became apparent that no one had examined the barriers or considered whether they should have.  Not even the forensics expert for the state, metallurgical engineer, had examined the barriers, considered their function, determined if they performed properly or whether they had been installed properly.  The state expert had also failed to examine the vehicles.  During cross examination, he testified he was never asked to visit the scene, trace vehicles, examine them or examine the roadside barriers – so he did not do any of this.  He wrote his report and formed an opinion solely on what he had received on paper.

While some photographs had been taken of the unfortunate young lady’s car at the scene, her vehicle had never been examined and photographed properly in daylight conditions.  Even the damage to her vehicle – and there was more than just the hole in the windshield – had not been examined or recorded and photographed properly.  There were only nine photographs of the scene, 13 of Lisa’s car and 13 of the Ford Fiesta – taken some time later at a different location – in the police docket.

Photo of the scene, taken by Stan Bezuidenhout

Bezuidenhout took 168 photographs in total of the scene, 280 of the Ford Fiesta and also made several video recordings at, around, and of the scene.

The driver of the Fiesta was ultimately charged with culpable homicide, reckless and/or negligent driving and inconsiderate driving.  He faced a long and arduous trial, lasting from 2008 to late in 2011.

Damage to Left of Lisa’s car – poor quality as original

But what does all this have to do with roadside barriers?

Bezuidenhout had been appointed in this matter by the accused and was immediately struck by the sheer absence of evidence.  There was vital evidence that was missing and glaring evidence that had been completely ignored by the prosecution.

It was clear to him that the prosecution had tunnel vision.  Only three elements had ever been explored by them:  They considered the driver of the Fiesta their primary and only suspect in the cause of the collision; they considered the damage to the barriers proof of his speeding, since he “was going so fast that he crashed through the barriers”; and they concluded Lisa had been killed by something kicked up by the crash, since it had been “so severe” in any case.

Comparison of marks

In spite of this, his speed had not been determined by the state-appointed expert, who testified that “the speed of the vehicles could not be determined.”  The prosecution had never even considered, explored, or investigated the performance of the barriers.

In trial, the state expert even testified that he had never examined the barriers and that – although he was a metallurgical engineer and although the barriers were made from metal – he was not “a barrier expert” and could thus make no comment on the barriers at the scene.

3D Model

The prosecution and their police investigators had never found and actually something from this crash under consideration.  The possibility that a stone or stones could have been thrown onto her vehicle as she passed under several bridges she might have encountered earlier, had never been considered.

The court had merely inferred this link from available evidence:  a singular small pebble, possibly the size of a person’s thumb, which had been found inside Lisa’s car, on her dashboard.  This pebble had never been examined for material and matched to the scene.  There was simply no way to conclusively prove that this pebble was not the only item that had entered Lisa’s car in the crash.  

Barrier design specifications

There was damage to the left side of Lisa’s car, in the shape of the long, narrow object that had never been explored or explained by the prosecution and their police investigators or their expert.  This was on the side of her car that was furthest away from the side where the Fiesta’s crash had occurred, so it could not have been caused during the events under consideration in this collision.

The driver of the sports car and the one who had allegedly seen the collision happen, admitted she liked to drive fast (using the car she drive then – a Maserati – in support of her boastful claim) and explained that debris from the Ford Fiesta and/or the barriers had flown onto her vehicle, causing damage to her bonnet and front windshield.

She could not explain why she had never photographed the damages she alleged had been caused to her vehicle, she could not explain why she had not stayed at the scene to give her details to police; and she denied seeing or being anywhere near the accused’s vehicle at anytime prior to the collision.

Other witnesses destroyed this version, since they testified consistently that they saw her vehicle some time earlier behind, and very close to the Ford Fiesta, seemingly “racing” against it. 

Bezuidenhout found trace evidence, in the form of marks at exactly the correct height, which would imply the Mercedes SLK would have made contact with the rear of the Ford Fiesta and that this could have made contact with the rear of the Ford Fiesta and that this could have resulted in a loss of control, since they would both have been going through the same bend when this could have happened.

While everyone was focusing on the “speeding vehicle”, Bezuidenhout started looking into the vehicle dynamics and the fact that the vehicle had been able to crash right through the barrier at all.

All versions had the accused driving at less than or around 100km\h in a car that weighted a mere 900-odd kilograms, yet his car had been able to crash “right through the barrier” at a very small angle – less that 5 degrees.  This made no sense since barriers are supposed to be designed to prevent exactly this.

There were a number of issues detected with regard to the installation of the barriers.  But in order for you to grasp the full extent and seriousness of these issues, you need to understand and grasp what properly installed barriers are supposed to look like and what their design capabilities are supposed to be.

Barriers are (typically) steel and wood dividers that are designed to prevent vehicles from departing from the roadway when they lose control or are involved in collisions.  In order fix this to be achieved, there are a number of physical and design specification that must be adhered to.

Barrier on Wooden Blocks

Many, if not most, barriers in South Africa- also called guardrails, roadside barriers, Armco barriers or physical barriers – are designed and built by Armco, in turn, works according to international specifications and adheres to SABS standards.  But it does not typically install all those barriers; the company often just specifies and supplies.  Contractors, typically appointed by municipalities or local authorities, more often than not do the physical installation and maintenance.  But this case has left Bezuidenhout with serious questions about the capacity and ability of some contractors to do so.

These are some of the design specifications that should be adhered to in order for an Armco Barrier to perform properly:

Section types

The Armco barriers used in most South African applications are of the W-type variety.  If you look at them from the end, they form a W-shape. This shape is considered effective, since it provides sprung resistance, linear stiffness, adequate distortion (flattening and expansion) under stress and reduced friction (W-face contact) for optimal redirection, buy not at the expense of overall strength.

In the case under consideration, the same type was installed.  The sections were of the correct size, design and length.  Unfortunately, this alone is not adequate for the guardrail to operate as intended.  In many installations, this is the only feature ever considered adequate.  The correct rail material is used – but without all the other specifications being adhered to, this system will simply not function effectively.

Steel strength and type

One cannot simply bend any steel into a W-shape, attach it to some posts and consider it a proper guardrail.  The thickness, hardness, steel type, physical characteristics and durability all feature in the design.  Armco barriers are built from the required material of the correct specification: 1mm thick mild steel. When examining the barrier at the scene where the collision occurred, Bezuidenhout found that the W-sections seemed to comply with all physical design requirements.  This meant that, if there were any issue with performance, it would not relate to the physical W-sections supplied by Armco.


According to the Armco specification, the center of the W-sections is supposed to be 720mm high (measured from the road surface).  This height is chosen as a function of the most likely height of the center of mass of vehicles that the system is supposed to retrain.  In this case, it was found that the height of the center of the W-sections range from 35cm to 1 meter from one side of the scene to the other, with vastly varying values for all posts in between.  This was a clear indicator with the system had not been properly installed.

The heights would easily compromise the design intention if a force was applied at the design height.  This means there would be a risk of a vehicle being able to ‘ramp over’ the barrier (where it is too low) or of the barrier intruding into the vehicle or the vehicle injecting underneath the barrier (if it was too high) – increasing injury and fatality risk to road users.

Post spacing

The Armco specification further requires that the posts be installed 3.81m apart for this installation.  The section length (post spacing) relates to the length of each W-section with its specified overlap, as required.

The spacing is specifically designed to ensure the system performs as designed.  It has to be strong enough to resist impacts, elastic enough to cause relapse (forcing a vehicle back onto the road) and fragile enough not to present a dangerously stiff obstruction causing impact injury or fatality.

Further, the system is not designed for direct assault (at 90 degrees), but rather to ‘deflect’ vehicles back onto the roadway, preventing them from crossing over onto approaching traffic lanes or down embankments or drops.

In the case under consideration, Bezuidenhout found that the post spacing for most posts was correct.  All the posts that were present were all 3.81m apart.  Some posts were found to be missing completely, effectively increasing the gap between adjacent posts to over 7.6m – well beyond the specified gap.

At those (missing post) positions, it was found that the barrier was suspended on rocks or wooden spacers packed loosely on top of each other, on the kerb.  This was coincidentally also found at the area where the Ford Fiesta “crashed through the barriers”.  This would completely have compromised the design specification and operational capacity of the guardrail system. 

Disturbingly – even sometime after this fatal collision – photographs revealed that the same situation was still prevalent in many locations around Cape Town and the rest of South Africa.

Post types and installation

Even the type of post used is considered in the design specification.  The wood needs to be of a specific thickness, length, hardness, and environmental resistance, and needs to be installed to 1.8m in length.  This means the posts need to be installed at a penetration depth of approximately 1.3m.

They also need to be installed into prepared earth.  The soil density must be of a specific specification.  This enables the posts to break at specific load values and not too early, while the ground must not be so hard as to cause them to break prematurely or so soft that they can simply wedge out or bend over completely.

In the case under consideration, the posts were found to have clearly been installed by laymen or by installers with no consideration for the specifications or with no training.  Some posts were simply inserted into the old holes of previous posts, which were obviously damaged in the many collisions that had occurred before, at the exact location.  In some cases, Bezuidenhout could fit almost his entire arm into the ground next to the posts.

Obviously, these posts would present almost no resistance to forces:  as observed and expected, they would simply ‘fall over’ under assault.  The lackluster installation would also cause them to pivot and eject debris when forces are applied – one of the possible causes of the debris that might have killed Lisa on that lovely Saturday afternoon.

Other posts were installed, and concrete or cement then laid around their bases.  This completely inhibits their ability to respond to applied forces before breaking under strain. Since they were now void of any ’play’ or deformation leverage, these posts would break under much reduced forces than the design specifications required.  Coincidentally, a lot of loose rocks, debris and cement or concrete pieces could also be seen lying around the bases of the posts.  During a collision, any of these would become a projectile if the system failed.

Fixtures (bolts used)

The integrity of the whole guardrail system relies on the smallest yet one of the most important components used on the installation:  the bolts that hold the W-section to the posts and those that join the sections together.

These bolts are not just any bolts.  They are of a specific design.  In order for you to know which system you are dealing with; the name of the manufacturer is typically displayed on top of the bolt head.  But again, the manufacturer is often not the installer – especially in municipal areas.

The bolts need to be of specific size and strength specification, and they need to be tightened to exactly 75Nm. This enables them to provide enough resistance to hold the system together in the event of an applied force, but at the same time they should allow the W-sections to expand along elongated mounting holes to just the correct amount and with enough friction to allow the system to effectively absorb energy and deform until it ultimately fails.

At each section joint, there must be six bolts – all of the same size and strength and all tightened to the same specification.  In the case under consideration, it was found that most of the bolts were seemingly hand-tightened while many mounting holes were found to have no mounting bolts at all.  This would compromise the guardrail system at its core.

So, what happened to the accused in the case?  He was found guilty of culpable homicide.  The court concluded he had been driving too fast, lost control and crashed through the barrier as a result of that speed.

While the metallurgical engineer appointed in the prosecution had never even bothered to look at the barrier, never examined any of the vehicles and never proved any of his conclusions, it causes concern that the court simply did not fully grasp the implications of the utterly poor quality of installation of the barriers at the scene.  But the trial continues even now in 2014 and the case has gone on appeal. 

Perhaps one day the court will grasp the true design goal of the guardrail systems on our highways:  to save lives.  Perhaps if installers are ultimately held accountable, there will be an improvement.  Until then, this case is a classic illustration that there is much yet to be learnt about crash cause analysis and even more to be done before the carnage on our roads will be properly addressed.

Have a look at the guardrails around you.  Are they properly installed?  Now what if you lost control in the rain and ‘crashed through’ a guardrail and into oncoming traffic?  If those rails are not properly installed and fail to perform as designed, would you also be found guilty of speeding and culpable homicide?  Perhaps…

Leave a Comment