Bitumen Sampling Procedure: Taking a Sample That Settles an Argument

Testing bitumen is a mature and heavily standardised activity. Penetration by ASTM D5, softening point by ASTM D36, flash point by ASTM D92 and solubility by ASTM D2042 represent more than a century of accumulated practice between them, and each carries a published precision statement that says how far two competent laboratories are allowed to differ. Sampling has standards too — ASTM D140 is the standard practice for sampling asphalt materials — but it is carried out on a hot tank top, in a warehouse aisle or beside a loading gantry, usually by whoever was available, and almost never with the discipline the laboratory applies afterwards.

That imbalance is where disputes come from. When a buyer’s arrival test disagrees with the seller’s Certificate of Analysis, only three explanations exist: the material changed between loading and testing, one of the laboratories is wrong, or the two results describe two different materials because they came from two unrepresentative samples. In practice the third is the common one, and it is the one nobody can disprove after the fact.

The sample is the only part of the cargo a test ever sees

A thirty tonne road tanker holds about thirty million grams of bitumen. A penetration test consumes a small fraction of that, poured into a container that ASTM D5 sizes at 55 mm diameter by 35 mm depth for penetration values below 200 dmm. Everything a buyer subsequently claims about thirty tonnes rests on a ratio of roughly one part in several hundred thousand. That ratio is defensible only if the sample was drawn so that every part of the cargo had a fair chance of being in it. A draw from the one working tap on a tank, or from the drums standing at the container door, does not meet that condition, and no amount of laboratory accuracy downstream rescues it.

Sampling error is directional, not random

Random error averages out with more measurements. Sampling error usually does not, because the shortcuts people take are correlated with the material they miss. The convenient tap sits at a fixed height, so it always samples the same layer. The drums at the door were the last filled, so they always come from the end of a run. The top of a cooled drum is an oxidised skin, so a scrape from the surface always reads harder than the drum. Each shortcut has a direction, and taking more samples the same way simply confirms the bias with greater confidence.

The evidence ends up asymmetric

Consider what each side holds once a claim is filed. If the sample was drawn at the load port by an independent inspector, split into sealed and numbered portions and lodged with both parties, the disagreement reduces to a test: break the seals in front of an agreed referee laboratory and read the answer. If it was not, the buyer holds material of uncertain provenance drawn from a tank he controls, and the seller holds a certificate from a laboratory the buyer does not accept. There is nothing to test, only two positions to argue and a commercial relationship to spend. That asymmetry is the whole reason sampling belongs in the contract rather than in an operating procedure nobody reads.

What a static tank actually contains

Bitumen conducts heat poorly and does not mix itself. With coils at the bottom and no circulation, a bitumen storage tank develops a vertical profile: hottest immediately above the coil bank, cooler through the middle, coolest against the shell and at the free surface under the roof, where a stiffened crust can form. Composition follows temperature and history. The bottom holds whatever settled — carbon shed from a fouled coil, sediment and any free water. The top has had the longest exposure to the air in the vapour space and is the most oxidised. If the tank was topped up rather than emptied between receipts, the layers are literally different cargoes. A modified binder adds a second mechanism: the polymer-rich phase is less dense than the bitumen around it and migrates upward, which is why ASTM D7173 exists as a separation test in the first place.

The depths the standards use

ASTM D4057, published jointly as API MPMS Chapter 8.1, defines the sample positions that inspection reports are written in. They are worth learning because they are the vocabulary of the argument:

• Upper sample — drawn at the middle of the upper third of the liquid, that is one sixth of the liquid depth below the surface.
• Middle sample — drawn at the middle of the tank contents, one half of the liquid depth below the surface.
• Lower sample — drawn at the middle of the lower third, five sixths of the liquid depth below the surface.
• All-levels sample — obtained by lowering a closed sampler as near as practicable to the outlet, opening it and raising it at a rate that leaves it about three quarters full as it emerges. It is not a true average, because it collects unequally with depth, but it is one sample that has seen the whole column.
• Running sample — the sampler is lowered and raised through the column with the opening uncovered throughout.
• Composite sample — upper, middle and lower blended in known proportions to represent the parcel as a whole.
• Outlet or drain sample — drawn from the outlet line. Valuable for finding water and sediment on the tank bottom, and misleading if used on its own as a cargo sample.

ASTM D140 applies the same principle to bituminous materials specifically: where material is held in a tank, samples are drawn at more than one level rather than from a single point, and where the tank carries sampling taps at different heights, the taps are used rather than one outlet. The working rule for a commercial cargo is to take the three depth samples, test them separately whenever there is any doubt about the tank, and composite them only when a single number representing the parcel is what you need.

Taps, samplers and dead legs

Two mechanical details ruin more tank samples than anything else. The first is the dead leg. A sampling tap holds a volume of stagnant material between the tank wall and the valve seat that has been sitting cooler and oxidising since the tap was last used, and on a hot bitumen tank it may be part solid. Draw that first and you have sampled the valve, not the tank. Flush the tap to waste until fresh product at tank temperature is running, then take the sample. A tap that cannot be flushed because it is blocked is not a sampling point.

The second is the sampler itself. A weighted sampling can or bottle lowered through the hatch must be clean, dry and warm enough that the sample does not chill against the container wall; a cold sampler dropped into bitumen at 160 °C collects a solidified shell that never enters the bottle. Cleaning a sampler with solvent and then using it before it has completely dried puts solvent into the sample, which lowers the flash point measured by ASTM D92 and shifts penetration by ASTM D5 in a way that looks exactly like a supply problem and is impossible to argue away afterwards.

Safety at the tank top

Sampling is one of the few routine reasons anybody stands over an open bitumen hatch, and it carries the two hazards that put people in hospital. Hydrogen sulphide (H₂S) evolves from hot bitumen and accumulates in the vapour space of tanks, tankers and tank containers. The ACGIH threshold limit value is 1 ppm as an 8-hour TWA with a 5 ppm short-term exposure limit, and the NIOSH IDLH value is 100 ppm. H₂S deadens the sense of smell at higher concentrations, so odour is not a warning system: gas test before opening, approach and stand upwind, never put your head over an open hatch and never sample alone. The second hazard is contact. Bitumen at storage temperature adheres to skin and keeps transferring heat after contact, so wear a full face shield over safety glasses, heat-resistant gauntlets with sleeves worn over the glove cuffs so nothing can run inside, and trousers outside the boots. Because a penetration-grade paving bitumen carries a specified minimum Cleveland open cup flash point of 250 °C by ASTM D92, with comparable minima elsewhere — 232 °C in ASTM D946 and 220 to 230 °C across the grades of EN 12591, so check which figure your own specification carries — the material at the hatch is a Class IIIB combustible liquid under NFPA 30 and the dominant risks during sampling are burns and gas exposure rather than static ignition. Bonding practice under API RP 2003 still applies at any terminal that requires it.

Water, and the products that are not paving grade

Two hazards belong to sampling specifically. The first is water. Free water settles to the bottom of a storage tank and sits under the heel, and water trapped beneath bitumen held far above the boiling point of water flashes to steam the instant it is disturbed, expanding by orders of magnitude and throwing hot binder out of an open hatch or out of a sample container. That is the reason a drain or outlet sample on a tank suspected of holding bottom water is taken slowly, standing clear and to one side of the discharge, and the reason a washed sample can must be completely dry before hot product goes into it. A trace of wash water in a can that then receives bitumen at storage temperature will boil under the surface and spit.

The second is the products that are not paving grade. Cutback grades contain solvent and behave nothing like the material described above. Their flash point is measured on the Tag open cup by ASTM D1310 rather than the Cleveland cup of ASTM D92, and ASTM D2027 sets a minimum flash point of only 38 °C for medium-curing cutbacks — near ambient in a hot climate, which means the vapour above the liquid can be inside its flammable range at the moment the hatch is opened. Cutbacks and emulsions are sampled cold, into a closed container, with no open flame and no heat applied to the sample at any point in the chain. Heating a cutback sample over a flame to make it pour is the single most dangerous thing anyone does with a sample container.

Why the drums at the door are not the consignment

A container is loaded in order, so the drums standing at the door are the last ones stuffed, and in most yards they are also among the last filled. They come from one part of one production run. If the run drifted, if a grade change happened mid-shift, or if the final drums were topped from a different tank, the door drums are precisely the units that will not show it. They are also the units a seller can most easily anticipate being inspected. Nothing about drawing from them is dishonest in itself; it is simply the least informative choice available, and it is the choice that gets made when the inspection is scheduled for an hour that does not allow anything else.

The only defence is random selection across the consignment, made before the container is stuffed, or made by devanning far enough to reach drums from more than one row and more than one tier. Both are slow and expensive, which is exactly why sampling at the filling line is the professional answer for packed goods.

Sample at the filling line, not the finished drum

A cooled drum of bitumen is a solid block with three different materials in it: an oxidised skin at the exposed top surface, a shell of material that cooled quickly against cold steel, and a core that cooled slowly over hours. None of the three is the drum, and a scrape from the top is certainly not the consignment. Sampling from the filling nozzle while drums are being filled resolves all of it at once. The material is hot, moving and homogeneous, the sample can be tied to a filling time and to a stated range of drum serial numbers, and nothing has to be cut open. Where a contract requires drummed material to be sampled, the sensible clause names the filling line as the sampling point and requires the drum numbers covered by each sample to be recorded on the label and on the inspection report.

Where a finished drum has to be sampled, ASTM D140 deals with solid and semi-solid material in packages by requiring the sample to be taken from below the exposed surface and clear of the container wall, so that the oxidised skin and the fast-cooled wall material are excluded. A figure commonly applied in inspection practice is at least 75 mm below the surface and at least 75 mm in from the wall, taken by cutting or coring; confirm the figure in the edition of the standard your contract names rather than relying on the convention. Coring a solid drum needs a proper tool and generates heat of its own, so the alternative — selecting one drum at random, melting it in full and sampling the melt — is often cleaner. It destroys one drum in order to describe the rest of them honestly.

Bags are harder, and the film is part of the product

A one tonne jumbo bag or a meltable poly bag is a single cast block with a worse surface-to-core history than a drum, and there is no route to the interior that does not open it. Two honest options exist: sample the filling line during production, or select a bag at random and melt the whole unit before sampling the melt. There is a second point that is easy to overlook on meltable packaging. The film is designed to melt into the binder and go into the mix with it, so a chip of bitumen taken from inside the bag does not describe the material that will actually enter the mixer. Where that matters to the specification — and it can matter to a solubility result by ASTM D2042, which is the line most often used to detect adulteration — test the material in the form in which it will be used, and write that into the sampling clause so it is not discovered as a surprise at the laboratory.

How many units to take

There is no universal number, and the number that governs is the one written into the contract. Two counting conventions are in general use across petroleum and general goods inspection, and both appear in the table below. The cube root of the number of packages, rounded up, comes from petroleum manual sampling practice and keeps the workload realistic on very large lots. The square root, rounded up, is more conservative and is common in general inspection work. Neither is a requirement of ASTM D140, and neither should be quoted as though it were. Whichever is adopted, three conditions matter more than the count: the units are selected at random rather than by convenience, the selected drum or bag numbers are recorded, and each unit is kept as a separate sample rather than blended, so that a single bad unit can still be identified instead of being diluted into an acceptable average.

What a proper retained sample set looks like

• Drawn at the same time, from the same increments, as the sample that is tested. A retained sample taken later, or from a different point, is simply a different sample and defeats the purpose entirely.
• Sealed at the sampling point with tamper-evident numbered seals, in the presence of an independent inspector where one is engaged, with the seal numbers recorded on the inspection certificate and cross-referenced on the Certificate of Analysis.
• Held by both parties. Buyer and seller each hold a sealed portion, and the inspector holds a third wherever possible. One party holding all the retained material is not a retained sample scheme; it is a courtesy that evaporates the moment it is needed.
• Retained for longer than the claim period written into the contract, with margin for notice, appointment of a referee laboratory and shipping the sample to it. If the contract allows claims for a stated period after discharge, retaining samples for exactly that period is not enough.
• Stored so that the sample does not change. Cool, dry, upright, out of sunlight and never reheated. Bitumen ages very slowly at ambient temperature, which is what makes a retained sample useful months later; heat cycles are what destroy it.

Why it settles disputes that nothing else can

When a claim arrives, the real question is never whose laboratory is better. It is which material was tested. With sealed retained portions in both hands, the parties agree a referee laboratory, break the seals in its presence and test. The result is accepted because both sides watched the same container being filled and sealed from the same increments. Without them, the buyer’s only evidence is material drawn from his own tank after discharge, and the seller can raise handling, heating, contamination with the receiving tank heel, or time in storage. Every one of those is a plausible alternative explanation, and the seller does not have to prove any of them — he only has to raise them.

Chain of custody is a document, not a sample

ASTM D4840 is the published guide for sample chain-of-custody procedures, and the principle it encodes is simple: from the moment of sampling to the moment of testing, one identified person is responsible for the sample at all times, and every transfer is signed and dated. A sample with a gap in that chain — left in a vehicle overnight, carried by an unnamed driver, received by a laboratory that did not record the seal condition — can be argued to have been substituted or tampered with. The argument does not have to be proved to work. It only has to be available.

Under ISO/IEC 17025, an accredited laboratory records the condition of the item on receipt and notes any departure from the specified condition. That requirement works in your favour: a report recording a seal number intact on receipt is worth considerably more than a report of the same test values with no such line. Ask for it in the instruction to the laboratory, and then read the report to confirm it is there.

When two laboratories still disagree

Assume the retained samples were sealed and tested properly and the two results still differ. That is not automatically a dispute, because every test method carries a published precision statement: repeatability for one operator with one apparatus on the same material, and reproducibility for two laboratories working independently. ASTM D3244 is the practice for using test data to determine conformance with a specification, and it uses that reproducibility figure to decide whether two results genuinely conflict or fall inside the method’s own noise, and to derive a single assigned value where they do not. Before escalating a small difference in penetration, look up the precision clause of ASTM D5 for the penetration level concerned, and the corresponding clause in ASTM D36 for softening point. A meaningful share of what gets raised as a quality claim sits inside the reproducibility of the method being argued over.

Load port beats arrival, and it is not close

Once a cargo has been discharged, pumped through a hose the buyer owns, into a tank the buyer heats, and held for a period the buyer controls, the supplier has a long list of legitimate questions about anything measured afterwards. Was the receiving tank empty before discharge? What was the heel, and what grade was it? At what temperature was the product held, and for how long? Was the coil covered throughout? Was the arrival sample drawn at one level or three? None of those questions has to be dishonest to be effective. They are the questions a competent engineer would ask, and they are unanswerable from the buyer’s side unless the receiving operation was documented as carefully as the loading.

Sampling at the load port under independent supervision removes the entire category. The material is sampled before it leaves the seller’s control, in the seller’s presence, by a party neither side employs. If the cargo is out of specification, that is established at a point where the seller cannot attribute it to the buyer’s handling — and, far more usefully, at a point where the shipment can still be stopped. A test on arrival tells you about a problem that has already cost freight, duty, demurrage, storage and a place in a construction programme.

None of this makes arrival testing pointless. It is the check on transit and on identity, and any buyer receiving bulk should sample the receiving tank before discharge as well as the delivery, so the heel is a known quantity rather than a later argument. But arrival testing is the second line of evidence, not the first.

What to write into the sampling clause

• The sampling standard and edition — ASTM D140 or EN 58, with EN 12594 or the equivalent for sample preparation, named alongside the test methods rather than instead of them.
• Who samples — an independent inspection company appointed and instructed jointly, not the seller alone and not the buyer alone. Where the company is not named, name the appointment mechanism.
• Where and when — at the load port, at the point of loading, during or immediately before loading rather than from storage several days earlier.
• Sampling points and increments — three depths for a static tank, three increments across a transfer for a moving stream, and a stated number of packages selected at random for drummed or bagged goods.
• Sample size and container type — enough material in each portion for a complete re-test, in new containers of the specified type and cleanliness.
• The number of sealed portions and who holds them — laboratory, seller, buyer, and inspector where one is engaged.
• Sealing and identification — tamper-evident numbered seals, with the seal numbers recorded on the inspection certificate and on the Certificate of Analysis.
• The retention period — expressed to run beyond the claim period in the contract, with a stated storage condition and a stated location.
• The referee laboratory — named, or selected by a stated mechanism, with the cost allocated by outcome.
• How disagreement is resolved — explicitly by reference to the reproducibility of the test method, following ASTM D3244, rather than by whoever shouts longest.
• The status of the inspection certificate — whether quality and quantity determined at the load port are final and binding between the parties. This is the clause that decides where the risk actually sits, and it deserves more attention than the price.
• Who pays for sampling, inspection and testing, and what happens to that cost if the cargo fails.

Independent inspection at the load port is one of the smallest line items in an export transaction, and the drawing, sealing and lodging of samples is most of what is actually being bought. A buyer who declines it to save that line, and then relies on an arrival test taken from his own tank, has exchanged a cheap piece of evidence for an expensive argument. What the inspector’s report should contain, and how it sits alongside the Certificate of Analysis, is covered on the documents pages rather than repeated here.