What actually governs a UAE construction project
Most disputes we are asked to look at begin with the same assumption: that the contract is the complete statement of the parties rights. On an onshore UAE project it is not. The UAE Civil Code supplies a body of mandatory and default rules that sit behind every construction contract governed by UAE law — obligations of good faith in performance, restrictions on the abusive exercise of a contractual right, principles governing the muqawala (contracting) relationship, and a power in the court or tribunal to review agreed compensation against actual loss suffered. None of this depends on the contract referring to it.
That matters commercially in a specific way. A liquidated damages rate agreed at tender is not automatically the number that will be awarded; a party facing an LD deduction can put the employer to proof of loss, and a party facing an inadequate LD cap may find the position is not as protective as the drafting suggested. Similarly, a right exercised strictly in accordance with the contract — a termination, a bond call, a rejection of a claim on a technicality — can still be attacked as an abuse of right if the surrounding conduct supports it.
The second layer is jurisdictional. A project physically located onshore, one located inside the DIFC or ADGM, and one governed by a foreign law with a UAE seat are three different legal problems. Free zone projects are subject to those zones own contract and civil law regimes and their own courts. Do not import an onshore analysis into a DIFC development, or the reverse.
FIDIC on UAE projects — and where UAE law overrides it
FIDIC is the dominant contracting framework here. The 1999 Red, Yellow and Silver Books remain the most frequently encountered, with the 2017 second editions appearing more often on newer infrastructure and utility work. Major employers — utilities, transport authorities, national oil and industrial entities — use heavily amended FIDIC-derived bespoke forms. The amendments are where the real risk allocation lives, and they are rarely neutral.
The amendments we see repeatedly, and which change outcomes: hardening of the notice and claim provisions into strict conditions precedent; deletion or dilution of the contractor entitlement to cost on employer-risk events, leaving time but no money; transfer of unforeseeable physical conditions and change-in-law risk to the contractor; capping or removing prolongation cost recovery; and modification of the dispute board mechanics so that a decision is neither binding nor readily enforceable in practice.
Where UAE law and the FIDIC text pull against each other, the drafting does not automatically win. A time bar that operates to extinguish a substantial entitlement can be tested against good faith and abuse of right principles, and the outcome is fact-sensitive rather than automatic. Practitioners who tell you the notice bar is always fatal, or that a UAE tribunal will always relieve against it, are both overstating. The correct position is that the strength of a time-bar defence turns on the precise wording, whether the employer had actual knowledge, whether the employer conduct induced or acquiesced in the failure, and whether the bar produces a disproportionate forfeiture. Serve the notice.
Decennial liability and why it cannot be drafted away
Decennial liability is the feature of UAE construction law that most surprises international contractors and, more often, their insurers. The Civil Code imposes on the contractor and the supervising designer a joint liability for total or partial collapse of a structure, and for defects that threaten its stability and safety, running for ten years from handover. It is a liability the Code characterises as one the parties cannot contract out of — provisions purporting to exclude or limit it are treated as void.
Three consequences follow that are commonly missed.
- Your liability cap may not do what you think. An overall cap of liability negotiated at 20 or 30 per cent of contract price does not answer a decennial claim. Nor does an exclusion of consequential loss. The cap governs the contractual claims; the decennial exposure sits outside it.
- The ten years is the exposure window, not the period in which to sue. Once a collapse occurs or a qualifying defect becomes apparent, a much shorter limitation period runs for bringing the claim. Employers who discover a structural defect and spend eighteen months in commercial correspondence can lose the claim while the ten-year window is still open.
- The designer is exposed alongside the contractor. Where the consultant supervised as well as designed, the liability is shared. This is why decennial claims almost always arrive as multi-party proceedings, and why consultant appointments need to be reviewed with the same care as the main contract.
Insurance is the practical answer, and it needs to be procured and maintained on the correct basis, with the policy period aligned to the statutory exposure rather than to the defects notification period in the contract. In the DIFC and ADGM, the applicable civil law regime is that of the free zone, and a decennial-style claim must be located in the law that actually governs rather than assumed by analogy to the federal position.
Extensions of time, delay and concurrency
Delay claims fail more often on method and evidence than on entitlement. Two questions decide them: was the delay event on the critical path at the time it occurred, and can that be demonstrated from contemporaneous records rather than from a retrospective reconstruction. Tribunals here are increasingly willing to reject an analysis whose baseline programme was never approved, whose progress updates were never issued, or whose logic links were revised after the event to produce the desired result.
The Society of Construction Law Delay and Disruption Protocol is the common reference point for methodology, and the choice among methods is itself a contested strategic decision. The methods are not interchangeable, and the difference between two credible analyses on the same facts is routinely measured in months of entitlement.
Concurrency is where UAE practice diverges from the English position most sharply. UAE law approaches shared causation through apportionment: where both parties contribute to a delay, a tribunal has the conceptual tools to divide responsibility rather than treat the contractor claim as wholly good or wholly bad. That is materially different from an approach in which a concurrent employer delay yields time but no cost, and it changes how a concurrency defence should be pleaded. It also means a contractor should not concede concurrency as fatal, and an employer should not treat it as a complete answer.
Disruption is a separate claim from delay and is regularly pleaded as though it were not. Disruption is loss of productivity, which can occur without any completion delay at all. It requires its own evidence — labour allocation records, output measurement, a defensible measured-mile or earned-value comparison. A disruption claim built on a global cost overrun and nothing else is close to unarguable.
| Method | What it does | Where it works | Where it fails |
|---|---|---|---|
| As-planned vs as-built | Compares the planned programme against what was actually built, as a factual narrative | Simple projects; limited records; where a full critical path model cannot be built | Does not establish causation on a critical path; readily attacked as descriptive rather than analytical |
| Impacted as-planned | Inserts delay events into the baseline programme to model their theoretical effect | Early prospective assessment of an EOT application | Ignores actual progress and contractor-culpable delay; rarely persuasive as a retrospective claim |
| Collapsed as-built | Removes delay events from the as-built programme to show what would have happened without them | Where the as-built record is strong and the baseline is weak or unapproved | Requires a defensible as-built critical path; highly sensitive to the logic the analyst inserts |
| Time impact analysis | Models each event at the point it occurred against the then-current updated programme | Contemporaneously updated programmes; event-by-event entitlement; concurrency assessment | Depends entirely on regular programme updates; expensive; unusable where updates were never issued |
| Windows / time slice | Divides the project into periods and analyses critical path movement within each | Long projects with shifting critical paths; apportioning shared responsibility | Data-intensive; window boundaries can be selected to favour a party and will be challenged |
Variations, change orders and the instruction that never came
The most common variation dispute in the UAE has nothing to do with valuation. It is whether the work was a variation at all, and whether it was instructed by someone with authority to instruct it.
Contracts define the instruction route narrowly — a written instruction from the engineer, in a specified form, sometimes counter-signed. Projects then run on site conversations, marked-up drawings, revised issued-for-construction packages and emails from junior consultant staff. The contractor builds what it is told to build. Twelve months later, the employer position is that the work was always within scope, that no variation was ever instructed, and that the person who asked for it had no authority.
The counter-arguments are available but they are evidential, not automatic. A pattern of the employer accepting and paying for similarly instructed work supports a course of dealing. Knowledge and acceptance of the varied work by the employer supports an argument that rejection of the claim is inconsistent with the employer own conduct. Where work has been performed and retained, unjust enrichment principles are engaged. Each of these is a harder case than simply holding a compliant written instruction.
On design-build and EPC forms the fight moves upstream, into the interface between the Employer Requirements and the Contractor Proposal. Where the two documents conflict and the contract does not establish a clear order of precedence, the dispute becomes one about the scope the contractor priced. That is a drafting problem created at tender and litigated at completion. The order of precedence clause is worth more attention at negotiation than most of what surrounds it.
Payment, suspension and the on-demand bond
The UAE has no statutory adjudication and no statutory payment regime of the kind operating in the United Kingdom, Singapore or Australia. A contractor facing non-payment has contractual remedies and general law remedies, and nothing that produces a binding interim decision within weeks. That absence shapes every payment dispute on a UAE project.
The contractual route is notice, then suspension, then — if the default persists — termination. Each step is a risk decision. Suspension exercised without strict compliance with the contractual preconditions converts the suspending party into the party in breach, and a wrongful suspension is a gift to an employer looking for grounds to call the performance bond and terminate. We advise on the sequence before the first notice goes out, not after the second one has.
Pay-when-paid arrangements in subcontracts are widespread and are frequently drafted more aggressively than the general law will support. Whether such a clause operates as a genuine condition of payment or merely as a timing mechanism depends on its wording and on how the governing law treats a provision that makes payment contingent on a third party act. It should never be accepted on the assumption that it will be read down.
On-demand bonds and guarantees. A true on-demand instrument is autonomous from the underlying contract: the bank pays against a conforming demand and the merits of the dispute do not enter into it. Injunctive relief against a call is available in principle, but the threshold is high and the usual route is to attack the beneficiary conduct rather than the bank obligation. The practical work is preventive — negotiating the demand mechanics, requiring a certified statement of default, capping the amount, building in reduction on milestones, and setting an expiry that is not open-ended. Once an unconditional demand has been made, the options narrow considerably and quickly.
Defects liability and the subcontract chain
The defects notification period in the contract and the statutory decennial exposure are different animals, and conflating them is a recurring source of loss. The contractual period governs the contractor right and obligation to return and remedy; expiry of that period does not close off claims outside it, and certainly does not close off a structural claim. Employers who treat the taking-over certificate as the end of contractor exposure, and contractors who treat it as the end of their own, are both wrong.
The commercially decisive issue in most defects disputes is back-to-back alignment. A main contractor carrying a ten-year structural exposure and a twelve-month subcontract defects liability period is uninsured against its own supply chain for the remaining nine years. That gap is created at subcontract award and discovered at the point of claim. The same applies to caps, notice periods, governing law, dispute forum and insurance obligations. Where the main contract and the subcontract dispute clauses do not match, a main contractor can find itself arbitrating with the employer and litigating with the subcontractor over the same facts, with no mechanism to join the two and a real prospect of inconsistent outcomes.
Supplier and equipment claims have their own difficulty: title, risk and acceptance testing frequently pass at different moments, and performance guarantees on plant are usually subject to their own regime of testing, retesting and liquidated damages that operates independently of the construction LDs. Those two LD regimes interact badly if nobody has checked.
Arbitration, court, and the expert evidence that decides it
Construction disputes of any size in the UAE are predominantly arbitrated, most often before DIAC or arbitrateAD, with ICC arbitration common on internationally financed work. The main reasons are practical rather than doctrinal: a tribunal can be constituted with construction expertise, proceedings run in English on English-language documents, and the volume of technical material can be managed procedurally in a way that onshore court process does not readily accommodate.
Onshore litigation remains the right or only route in defined situations — where there is no arbitration agreement, where a decennial claim is brought against parties who are not all bound by the same clause, and where urgent relief is needed. Precautionary attachment over assets, and the preservation of site evidence before it is built over or removed, are matters where the onshore courts are frequently the effective forum whatever the contract says about arbitration.
The DIFC and ADGM courts are a genuine option where the parties and the project connect to those jurisdictions, and their procedures on disclosure and expert evidence will feel familiar to common-law practitioners. They are not a general alternative to the onshore courts for an onshore project.
On expert evidence: construction arbitrations are decided by experts more than by advocates. Delay, quantum and technical experts should be engaged early — before the claim is pleaded, so that the pleaded case matches what the records can actually support. Two failures recur. The first is instructing an expert after the case theory is fixed, which produces a report that has to argue rather than analyse. The second is a mismatch between the delay expert methodology and the quantum expert loss model, so that the two reports are internally inconsistent and opposing counsel does not have to work hard to say so.
Where this goes wrong — the failure modes we see repeatedly
These are not hypotheticals. They are the recurring patterns behind the claims that come to us too late.
- The notice that was never served because the relationship was good. The commercial team keeps the claim out of formal correspondence to protect the relationship. The relationship then deteriorates, and the employer relies on the time bar for everything that happened in the preceding eighteen months.
- The programme that was never updated. No approved baseline, no monthly updates, no as-built record. The delay analysis has to be constructed retrospectively, and the tribunal treats it as advocacy.
- Verbal instructions from the site. Work is built on the strength of a conversation and a marked-up drawing. The formal position at the end is that no variation was ever instructed.
- Liability caps assumed to cover structural exposure. Negotiated hard, priced into the tender, and irrelevant to a decennial claim.
- Suspension exercised one step early. The preconditions were not met. The suspending party becomes the defaulting party, and the bond is called.
- Subcontracts that are not back-to-back. Different defects periods, different caps, different dispute forums. The main contractor absorbs the difference.
- Global claims. A single lump sum for delay, disruption and loss of productivity with no attempt to link cause to effect. Tribunals here, as elsewhere, are unsympathetic where the records existed to do better.
- Experts instructed after the pleadings. The case theory is already committed. The expert either supports something the records will not carry, or contradicts the pleaded case.
- Sitting on a discovered structural defect. The ten-year window is open, so the employer takes its time commercially. The much shorter limitation period from discovery expires while the correspondence continues.
Every item on that list is preventable, and none of them is prevented at the dispute stage. The intervention that changes the outcome is a contract administration review while the project is running — checking that notices are being served, records kept, programmes updated and instructions documented. It costs a fraction of the claim it protects.
Frequently asked questions
Does UAE law override what our FIDIC contract says?
On an onshore project governed by UAE law, yes in part. The Civil Code supplies mandatory and default rules that operate whether or not the contract refers to them — good faith in performance, restrictions on abuse of a contractual right, the muqawala regime, decennial liability, and a power to review agreed compensation against the loss actually suffered. The contract governs most of the relationship, but it is not the whole of it, and a risk allocation that reads cleanly on the page can operate differently once the governing law is applied.
Can we exclude decennial liability by agreement?
No. The Civil Code treats provisions purporting to exclude or limit decennial liability as void. Your negotiated liability cap, your exclusion of consequential loss and your contractual defects notification period all govern the contractual claims; none of them answers a decennial claim for collapse or for a defect threatening the stability and safety of the structure. The practical response is insurance procured on the correct basis, with the policy period aligned to the statutory exposure rather than to the contractual defects period.
We missed the notice deadline. Is the claim dead?
Not necessarily, and not automatically alive either. The strength of a time-bar defence under UAE law turns on the exact wording of the provision, whether it is drafted as a true condition precedent, whether the employer had actual knowledge of the event, whether the employer conduct induced or acquiesced in the failure to notify, and whether strict application would produce a forfeiture disproportionate to the breach. Anyone who tells you the answer without reading the clause and the correspondence is guessing. Serve the notice; argue about relief only if you have to.
How is concurrent delay treated in the UAE?
Differently from the English position, and this matters. UAE law approaches shared causation through apportionment — where both parties contribute to a delay, a tribunal has the conceptual basis to divide responsibility rather than deliver an all-or-nothing outcome. Contractors should not concede that concurrency defeats a cost claim, and employers should not plead it as a complete answer. How apportionment is actually applied depends on the evidence and the tribunal, which is another reason the delay analysis has to be built on contemporaneous records.
Can we get an injunction to stop a performance bond being called?
In principle yes, in practice rarely and only on a high threshold. A true on-demand instrument is autonomous from the underlying contract: the bank pays against a conforming demand and the merits of the construction dispute do not enter the analysis. Relief is usually sought against the beneficiary conduct rather than the bank obligation, and the application has to be brought before the demand crystallises. The effective work is done at negotiation — certified statements of default, capped amounts, milestone reductions and a defined expiry.
Is there statutory adjudication in the UAE like in the UK?
No. There is no statutory adjudication regime and no statutory payment framework of the kind operating in the United Kingdom, Singapore or Australia. A contractor facing non-payment has its contractual remedies — notice, suspension, termination — plus general law claims, and nothing that produces a binding interim decision in weeks. Contractual dispute board mechanics under FIDIC can help, but their practical value depends heavily on how the employer has amended them and on how readily a decision can be enforced.
Should our construction dispute go to arbitration or the onshore courts?
For any substantial dispute, arbitration is usually the better forum: a tribunal with construction expertise, English-language proceedings, and procedures that can absorb large volumes of technical material and expert evidence. The onshore courts are the right route where there is no arbitration agreement, where a decennial claim involves parties who are not all bound by the same clause, or where you need precautionary attachment or urgent preservation of site evidence. DIFC and ADGM courts are a real option where the parties and the project genuinely connect to those jurisdictions, not a general substitute.
When should we bring in a delay expert?
Before the claim is pleaded. The most damaging sequence is to fix the case theory first and instruct the expert afterwards, which produces a report that has to argue rather than analyse and a pleaded case the records will not support. Engage the delay and quantum experts together at the outset, so that the methodology and the loss model are consistent with one another and with the documents you actually have. If the records will not support the claim you want to make, it is far cheaper to learn that before the pleading than during cross-examination.