Avoiding Fraud in Bitumen Trade

Fraud in bitumen trade rarely looks like fraud. It looks like a routine message at a busy moment, from a name you recognise, about something ordinary. The single most expensive email a procurement department will ever open is not a threatening one. It is a polite note explaining that the bank account has changed and apologising for the inconvenience.

Why this product attracts it

Four features of the trade do the work.

• Value is concentrated. A single container load or a single bulk parcel carries enough value to justify a serious effort, and the payment usually moves as one transfer.
• The product is fungible. Bitumen has no serial numbers and no visible identity. Two drums that look identical can be two different grades, and one of them can be a blend of things you did not buy.
• Quality is invisible without a laboratory. You cannot tell a 60/70 from an 80/100 by looking at it, and you certainly cannot see a failed thin film oven test. By the time the plant tells you, the cargo is discharged and paid for.
• Jurisdiction is split. Buyer, seller, bank, carrier and cargo are frequently in four or five different countries. That makes recovery slow, expensive and often uneconomic relative to the sum lost, which is precisely the calculation the fraudster is relying on.

Two families, and they need different defences

The first family is fraud by an outsider impersonating a real relationship. Nobody at either company is dishonest; a criminal has inserted themselves into the correspondence, usually by compromising one of the two mailboxes or by registering a domain that differs from the real one by a single character. The cargo is genuine and the seller is genuine. Only the bank details are wrong. The defence here is entirely procedural and it has nothing to do with the goods.

The second family is fraud by the counterparty itself: the advance payment against a cargo that does not exist, the cheaper grade shipped against a firm specification, the short weight, the reconditioned drum, the certificate that was never issued. Here the defence is contractual and evidential. It lives in how the contract is written, who inspects, what the documents must say and when the money is allowed to move.

The two structural principles

Everything on this page is an application of two ideas.

• Never let one channel both create an obligation and confirm it. If an email creates a payment instruction, an email cannot verify it. If a seller supplies the cargo, the seller cannot be the only party who tests it. Verification has to arrive by a route the other side does not control.
• Money moves as performance is evidenced. Every payment structure in this trade can be ranked by how closely it ties the release of funds to something that has demonstrably happened. That ranking is the whole subject of the payment sections below.

When it happens

Fraud clusters at two moments: the moment a payment is due, and the moment documents are presented. Those are the two points at which a small delay is most inconvenient and therefore the two points at which people skip checks. Build your controls so that the check is automatic at exactly those moments rather than something a person has to decide to do under pressure.

This applies to us as well

Everything below applies to this company as much as to any other supplier you approach. We would far rather you verified our details on a channel you already held, insisted on a batch Certificate of Analysis with the test methods named, appointed your own inspector at the load point, and structured payment so it follows performance, than that you trusted an email because it looked professional. A supplier who treats that process as an insult is telling you something useful. Nothing on this page is legal, banking or tax advice; contracts, remedies and reporting obligations are specific to your jurisdiction and to the wording you sign.

Invoice interception, and why it works

The mechanics are simple. Someone obtains sight of the correspondence between buyer and seller — most often by compromising a mailbox at one end, sometimes by registering a domain that differs from the real one by a single character. Substitutions that survive a quick glance include rn for m, l for I, a doubled letter, a transposed pair, or the same name under a different top-level domain. The criminal then waits. They do not intervene early, because early intervention exposes them. They wait until a real invoice is issued against a real cargo, and then they send a message advising that the account has changed.

What makes it effective is that every other detail is correct. The cargo is real, the invoice number is real, the tonnage is real, the tone matches the previous fifteen messages because the previous fifteen messages were read. In the worst version the message arrives from the genuine mailbox, inside the genuine thread, because the mailbox itself is compromised. In that case there is no spelling to spot.

The protocol that defeats it

Because the message may be indistinguishable from a real one, the defence cannot rely on detecting a fake. It has to rely on the fact that the criminal does not control any channel except the one the message arrived on.

• Fix banking details once, in the signed contract, and treat them as a contract term. A change to a payment instruction is then a contract amendment, not an email.
• Verify out of band, on details you already held. Call the number in the contract or in your own supplier record — never the number in the signature of the message proposing the change, and never a new number offered in that thread. Ours is +971 56 144 5733, it is published on this site, and it is the number to use if you ever receive a message claiming our account has changed.
• Speak to a person you have spoken to before, and ask them to state the account details to you rather than confirming details you read out. A confirmation question that can be answered yes is not a verification.
• Send a test transfer. A small amount first, confirmed as received by voice, before the balance moves. The delay is a day; the alternative is the whole payment.
• Look at the headers, not the display name. Check the actual sending domain, the reply-to address and the return path, and whether the message passed SPF, DKIM and DMARC. A reply-to that differs from the sender is the single most common technical tell.
• Make the rule apply in both directions. Tell your suppliers, in writing, that you will never change your own banking details by email either. A rule only one side knows about is not a rule.
• Give juniors permission to delay. Most of these frauds succeed because urgency was manufactured and somebody did not feel able to hold up a payment to make a phone call. Write it into the procedure so that no individual has to take that decision personally.

If the transfer has already gone

Speed is the only variable you still control. Contact your bank immediately and ask for a recall to be raised with the beneficiary bank; the realistic window is hours, not days, because the funds are moved onward and split as soon as they land. Report it to the police or the relevant national cybercrime or fraud reporting body, because a formal report is often what the receiving bank needs before it will freeze anything. Tell your counterparty at once, on the phone, since one of the two mailboxes is compromised and the other party needs to assume theirs is the one. Preserve the original messages with full headers rather than forwarding them, because forwarding destroys the evidence you will need. Then change credentials and enable multi-factor authentication on both mail systems before resuming any correspondence about payment.

Advance-fee fraud

The second attack asks for money that should never have been sent. Its signature is a demand for full payment in advance from a counterparty you have no history with, usually decorated with a price below the market and a deadline that has nothing to do with logistics: an allocation about to lapse, another buyer waiting, a loading slot that must be confirmed today.

Two things are worth understanding about it. First, the below-market price is not a mistake or a generous introduction; it is the payload. It is what suppresses the buyer’s scepticism and it is why the offer exists. In a market where freight, refinery economics and packing costs are visible to everyone, a genuine seller has very little room to be dramatically cheaper than the rest of the market for the same specification, and none at all to be cheaper without a reason they can explain. Ask what the reason is, and listen to whether the answer is a fact about the material or a story about the seller.

Second, the request escalates. Once a first payment is made, further payments appear: a performance bond, a port or authority fee, a document release charge, an inspection deposit, a demurrage guarantee. Each is small compared with the sum already committed, which is exactly the psychological lever being pulled. The correct response to the first unbudgeted fee is to stop, not to pay. Every legitimate cost in an export shipment either sits inside the price or is allocated between buyer and seller by the Incoterms 2020 rule you agreed, and a seller who is close to the cargo can tell you which of those two it is without hesitating.

What a defensible payment structure looks like

You do not need to refuse every advance. You need the advance to be a sum you can lose, and everything above that sum to be tied to evidence. A workable first transaction with an unfamiliar counterparty usually looks like one of these: an irrevocable documentary credit at sight, payable against a document set that includes an inspection certificate from a company you named; or a documentary collection where the bank releases the transport document only against payment; or a modest advance with the balance against shipping documents. If a genuine advance payment guarantee from a bank you can independently verify is offered, that changes the calculation — but verify it through your own bank, bank to bank, and never from a document that arrives as an attachment.

The autonomy principle

Documentary credits are governed by the ICC Uniform Customs and Practice for Documentary Credits, UCP 600, incorporated by reference into the credit itself. Three articles do most of the work.

• Article 4 establishes that a credit is by its nature a separate transaction from the sale contract on which it may be based, and that banks are in no way concerned with or bound by that contract even if a reference to it is included in the credit.
• Article 5 states it directly: banks deal with documents and not with goods, services or performance to which the documents may relate.
• Article 34 removes the last hope: a bank assumes no liability or responsibility for the form, sufficiency, accuracy, genuineness, falsification or legal effect of any document.

Read together, those provisions mean that a beneficiary who presents a set of documents that appear on their face to comply gets paid, and the bank neither knows nor is required to care what is in the container. Article 14(a) confirms that examination is on the basis of the documents alone. A credit is therefore protection against a seller taking your money and producing nothing at all. It is not protection against a seller producing the wrong thing and papering it correctly.

What a credit is genuinely good at

Having said all that, a credit is still far better than an advance transfer, and for reasons worth being precise about. It removes the possibility of your funds being taken before any performance whatever. It compels a document set to come into existence, which means a carrier, a chamber of commerce, an insurer and potentially an inspection company have all had to do something real. It inserts banks into the chain, which imposes deadlines and creates a record. And it gives you a lever: if any presented document is discrepant, payment is not automatic, and you have a decision to make rather than a loss to absorb.

Making the credit do more work

The way to close the gap between documents and goods is to require documents a fraudster cannot conveniently manufacture, and to specify them tightly.

• Name the inspection company and the certificate content. This is the highest-value clause in the whole credit. Beware of the drafting trap in Article 14(f): where a credit requires a document other than a transport document, insurance document or commercial invoice, and does not stipulate by whom it is to be issued or what data it must contain, banks accept the document as presented provided its content appears to fulfil the function of the document required. A credit calling loosely for a quality certificate can therefore be satisfied by a certificate the seller wrote about their own cargo.
• Require the Certificate of Analysis to state a batch number and the test methods, and require the batch number to appear on the packing list and the inspection certificate too. Data across documents need not be identical under Article 14(d), but it must not conflict — so a batch number that appears in one document and contradicts another is a discrepancy you can act on.
• Require a full set of original bills of lading, and consider consigning to the order of the issuing bank so that title is not released loose into the chain.
• Have the inspector report to you directly as well as issuing the certificate for presentation. A report that reaches you only through the seller’s hands is a report the seller has had an opportunity to edit.

Soft clauses cut in both directions

A soft clause is a condition in the credit whose satisfaction is controlled by the other party. The classic version harms the seller: a credit requiring an inspection certificate signed by the applicant, or a document issued by a party the applicant controls, means the applicant can withhold compliance at will and the beneficiary ships against a credit that can never be drawn. Buyers should understand this from the other side too, because a credit containing such a clause is evidence of bad faith wherever it comes from, and because a seller who spots one will rightly refuse to ship. Read the draft credit line by line before it is issued, and refuse any condition that neither party can independently satisfy.

Deadlines that catch honest parties as easily as dishonest ones

UCP 600 sets hard timing rules and they are worth knowing on both sides of a transaction. Article 14(b) gives each bank a maximum of five banking days following the day of presentation to determine whether a presentation complies. Article 14(c) requires a presentation including one or more original transport documents to be made no later than 21 calendar days after the date of shipment, and in any event not later than the expiry date of the credit. Article 16 governs refusal: a single notice of refusal, stating each discrepancy, given by the close of the fifth banking day. Most rejected presentations in this trade are not fraud; they are honest sellers who missed a date or misdescribed the goods against the credit.

The fraud exception, and why it is a poor safety net

Courts in most jurisdictions recognise an exception to the autonomy principle where there is clear evidence of fraud, and will in principle restrain a bank from paying. UCP 600 itself does not address fraud; it is entirely a matter of the applicable national law. In practice the exception is a weak protection for a buyer: the evidentiary standard is high, the relief must be obtained before payment is made, the timetable is measured in days, and the outcome depends on the courts of a jurisdiction that may not be yours. Treat it as a remedy that occasionally rescues a case, not as a control you can plan around.

Where collections and guarantees fit

A documentary collection under URC 522 is cheaper and lighter than a credit. On documents against payment terms the collecting bank releases the transport document only against payment, which keeps the seller in control of title until money moves — but no bank undertakes to pay, and the buyer still pays before inspecting. Demand guarantees under URDG 758 and standby credits under ISP98 serve a different purpose: they pay on a compliant demand, so they are how a buyer protects an advance, or how a seller protects against non-payment. Their weakness is that they are the instruments most commonly counterfeited. Any guarantee, standby or proof-of-funds document should be authenticated by your own bank through interbank channels, and an unsolicited offer to provide one is itself a warning sign.

Name the grade and the standard, not just the grade

A contract that says Bitumen 60/70 and nothing else has described a penetration band and left everything else open. A cargo can sit inside 60–70 dmm by ASTM D5 and still fail solubility, fail retained penetration after the thin film oven test, and carry a flash point that says solvent has been present. The contract needs the grade, the governing standard and the edition: for example penetration-graded binder to ASTM D946, or paving grade bitumen to EN 12591, or viscosity grade to IS 73:2013, or performance grade to AASHTO M320 or ASTM D6373. Standards are revised, so name the edition you mean or state that the current edition at the date of shipment applies.

Then attach the full specification as a schedule, with a minimum and a maximum for each line and the test method beside it. Substitution survives in the space between what was named and what was tested. Close the space.

Say how quantity is determined, not just what it is

Short weight is the quietest fraud in the trade because it does not fail, it just underdelivers. The contract should state four things: the quantity, the tolerance, whether the basis is net product weight or gross, and by what method the weight is established.

• Drummed cargo. Net weight is product only, with the drum tare excluded. State the nominal net per drum and the drum count. State that drums are marked with net weight and tare weight, and require weighing of a random sample of filled drums at load in the inspection scope.
• The container cross-check. Under SOLAS Chapter VI, Regulation 2, as amended and in force since 1 July 2016, the shipper must declare a verified gross mass for every packed container before it is loaded, established either by weighing the packed container or by weighing all packages and adding the container tare. That declared figure is a number you can reconcile against: verified gross mass, less the container tare marked on the door under ISO 6346, less the total drum tare, should land close to the invoiced net product weight. A material gap is a question worth asking before payment.
• Bulk cargo. Quantity comes from gauging and calculation, not from a declaration. Manual gauging follows API MPMS Chapter 3, temperature determination follows API MPMS Chapter 7, and the calculation follows API MPMS Chapter 12. Volume must be corrected to a base temperature using ASTM D4311 with a density measured to ASTM D70. This matters commercially: a volume measured at loading temperature and converted to tonnes using a density at 15 °C overstates the weight, because the hot volume is larger than the same mass would occupy cold. Insist that the temperature, the density and the correction are all shown on the survey report, not just the final tonnage.
• The seal. Record the container seal number on the bill of lading, the packing list and the inspection certificate, and require them to match. Seals for freight containers are classified under ISO 17712, with high-security seals designated H. A seal number that does not match across documents, or a seal that is intact but of a different type from the one certified, is a finding.

Describe the packing, because reconditioned drums are a real order

Reconditioned drums are a legitimate product with a legitimate market. They become fraud when they are supplied against an order for new ones. The contract should say new, unused and non-reconditioned in those words, and then add the detail that makes the words checkable: the steel thickness for body and heads, whether the interior is lacquered or plain, closure type, and the marking required on each drum — grade, batch number, net weight, tare weight and production date.

The physical tells at destination are consistent. Look for dents, ovality or distortion at the top and bottom chimes, which is where a drum is gripped and dropped. Look for fresh paint applied over old markings, visible as raised ghost lettering under the coating, and for rust bloom coming through new paint. Look for closures and lids of a different manufacture from the drum body, mismatched colours between lots, inconsistent empty weights, and any residual odour of a previous product. A single reconditioner’s stamp anywhere on a batch settles the question. Where drums carry UN packaging marks — which paving bitumen at ambient temperature usually does not, since it is not consigned as dangerous goods — the UN Model Regulations require a reconditioned packaging to carry the reconditioner’s own mark including the letter R, so on UN-marked drums the evidence is stamped into the steel.

Put drum condition and marking explicitly into the pre-shipment inspection scope, with dated photographs of the drums as filled and as stuffed. An inspection scope that covers quality and quantity but is silent on packing is the scope under which this fraud passes.

Sampling, retained samples and the right to retest

Testing is only as good as the sample. Sampling of asphalt materials is covered by ASTM D140, AASHTO T 40 and EN 58, and the contract should name the practice to be used. For drummed cargo, samples must be drawn across the load rather than from whichever drum is nearest the container door. For bulk, sampling should follow the loading operation rather than being a single grab. How many packages to draw, where in a tank to draw from, and how a sample is sealed and dispatched are set out in full on the sampling procedure page; the point here is contractual. A contract that does not name a sampling practice has left the most contestable step in the whole chain to whoever is holding the tin.

Then require sealed retained samples, taken by the inspector in the presence of both parties’ representatives, split and held by both sides for an agreed period after discharge. This single clause converts a future argument from an exchange of assertions into a laboratory question. Without a retained sample from the load point, a quality dispute discovered at the plant is unwinnable, because the seller will say the material was contaminated or aged after delivery and nobody can prove otherwise.

Set the remedy before you need it

State what happens if a result is out of specification: rejection, a price adjustment on a stated scale, or replacement, and who bears the cost of the retest. Name the laboratory or the mechanism for choosing one. And name ASTM D3244, the practice for using test data to determine conformance with specifications, as the route to resolve a difference between the seller’s result and yours. D3244 uses the precision statement published in the test method itself to combine two laboratories’ results into an assigned test value, which turns a stand-off into a calculation. A counterparty who refuses to accept a published resolution practice is telling you what they expect the result to be.

A COA is a test report, not a specification sheet

The distinction is the whole subject. A specification is a set of limits that applies to a product in general. A Certificate of Analysis is a record of what a laboratory measured on a specific quantity of material at a specific time. If a document shows the same numbers for every shipment, it is a specification with a new date on it, whatever the heading says.

The eight things a genuine report carries

• A batch or lot number that identifies the production run, and that also appears on the drums, the packing list and the inspection certificate. A COA with no batch number cannot be tied to anything you received, which means it evidences nothing.
• A sampling date and a sampling reference, ideally naming the practice used — ASTM D140, AASHTO T 40 or EN 58 — and who drew the sample.
• Test dates, separate from the issue date. These should be chronologically sensible: after production, after sampling, before or around loading. A thin film oven test takes five hours at 163 °C under ASTM D1754; a rolling thin film oven test runs 85 minutes under ASTM D2872. A report showing sampling, a full ageing test suite and loading all on the same day deserves a question.
• The test method beside every value, by designation. Penetration by ASTM D5, softening point by ASTM D36, flash point by ASTM D92, solubility by ASTM D2042, and so on. A number with no method is not a result.
• Correct units. Penetration in dmm, meaning tenths of a millimetre, not millimetres. Ductility in cm. Softening point in °C, labelled. Viscosity in the unit the specification actually requires — poise, centistokes or Pa·s are not interchangeable and a report that swaps them has not been read by anyone technical.
• Measured values that scatter. See below; this is the strongest single signal on the document.
• An identified laboratory, with a name, a physical address and a contact route.
• A named signatory with a role, not an unattributed stamp.

Values that scatter, and values that do not

Real measurement is noisy. Every ASTM and EN test method carries a precision and bias section giving repeatability — the spread expected between two results by the same operator on the same apparatus — and reproducibility, the spread expected between laboratories. Look up the precision statement for the methods that matter to you, because it tells you how much scatter is normal for that test, and therefore how much is suspicious.

The tell is simple. A report on which every value sits exactly on a specification limit was copied from the specification. A penetration reported as exactly 60, a softening point of exactly 49 °C, a flash point of exactly 250 °C, a solubility of exactly 99.0 % and a ductility of exactly 100 cm did not come out of a laboratory; those are the limits, in order. Genuine results land in the middle of bands as often as at the edges, they carry inconvenient decimals, and they vary from batch to batch. Related to this, ASTM E29 is the published practice for using significant digits to determine conformance, and it governs how a measured value is rounded before it is compared with a limit. A report whose figures are all reported to a coarser precision than the method produces has been simplified by somebody, and simplification is where invention hides.

Check the internal arithmetic

Some lines on a COA are computed from other lines, and computed lines can be checked without any equipment at all. The clearest example is retained penetration after the thin film oven test: the report should show the original penetration, the penetration of the residue, and the retained percentage, and the third should follow from the first two. ASTM D946 requires a minimum of 52 % retained penetration for grade 60-70. Export specifications commonly carry the same measurement expressed the other way round — a maximum 20 % drop in penetration after heating — and it is worth being clear that these are not the same limit. A maximum 20 % drop means a minimum 80 % retained, which is far stricter than the standard requires. Read your contract to see which of the two it names, because a cargo can satisfy ASTM D946 and still breach a contract written to the export line. If the retained percentage is stated with no residue penetration behind it, or the arithmetic does not reconcile, that line was written rather than measured. This pair of values is simultaneously the best predictor of premature pavement cracking and the part of the report most often absent.

The penetration index cross-check

Penetration at 25 °C and ring-and-ball softening point are not independent quantities. For a given bitumen they are linked by its temperature susceptibility, and that relationship gives you a consistency check you can run on any COA with a calculator. The penetration index, from the Pfeiffer and Van Doormaal relation, is computed by first finding the susceptibility A from the penetration P in dmm and the softening point T in °C:

• A = (log 800 – log P) / (T – 25)
• PI = (20 – 500A) / (1 + 50A)

Worked example: a COA reporting P = 65 dmm and T = 49 °C gives A = 0.0454 and a penetration index of about -0.83. That is entirely normal. Straight-run paving bitumens typically compute to a penetration index between roughly -1 and +1. Blown or oxidized grades sit far higher: an oxidized 85/25, with a nominal softening point near 85 °C against a penetration near 25 dmm, computes to about +3.3, which is exactly what its manufacturing route implies.

The use of this is diagnostic. If a document sold to you as a straight-run paving grade reports, say, 65 dmm against a softening point of 62 °C, the penetration index comes out above +2 — a blown-grade signature on a paving-grade certificate. Either the material contains blown or blended component, or the two numbers were invented separately by someone who did not know they were related. Be clear about the status of this check: the penetration index is an engineering consistency indicator, not a requirement of any specification, and a value outside the usual range is a reason to ask a question, not a basis for rejection on its own. Its value is that it is free, instant, and impossible to anticipate if you are fabricating numbers.

The laboratory has to be a real, contactable laboratory

A refinery’s own laboratory issuing its own COA is entirely normal and is not a red flag by itself. What matters is that the laboratory can be identified and reached. Ask for the laboratory’s name, address and a direct contact, and then use them. Where the value at risk justifies it, require testing by a laboratory accredited to ISO/IEC 17025, which is the international standard for the competence of testing laboratories and which governs how a test report identifies methods, items tested, sampling and test dates, and measurement uncertainty where relevant. Accreditation certificates carry an accreditation number and a scope, and the scope matters: a laboratory can be accredited for tests that have nothing to do with bitumen. Verify the accreditation in the public register of the accreditation body that issued it, and check that the body is a signatory to the ILAC mutual recognition arrangement.

Note the difference between two things that look similar. An ISO/IEC 17025 accreditation is about the competence of a laboratory to perform named tests. An ISO 9001 certificate is about a management system and its defined scope, and says nothing whatever about product conformity. Both are worth verifying — accredited management system certifications can be checked in the International Accreditation Forum’s public CertSearch database — but neither should be read as a statement about the cargo.

When the numbers look right and you are still not sure

Then buy an independent test. Have your inspector draw a sealed sample at the load point under ASTM D140 or EN 58, send it to a laboratory you chose, and compare. The cost is modest against the value of a cargo, and the result is decisive in a way that no amount of document analysis can be. Reserve the strongest version of this for the first transaction with any counterparty, and for any transaction where the price is meaningfully better than the rest of the market.

• The documentary credit — what a letter of credit does and does not protect against, the discrepancies that stop payment, and how to read one before you accept it