AN award that finally decides all issues brought before an arbitral tribunal is final and binding on the parties and exhausts the tribunal’s mandate. What happens, however, when proceedings are bifurcated and certain issues are finally decided in a first award, only for the tribunal to revisit them in a second award? Can this second award be annulled or is it merely another exercise of the tribunal’s authority, insulated from court review?

These questions were considered by the High Court of Australia in CBI Constructors Pty Ltd. v Chevron Australia Pty Ltd. [2024] HCA 28. By confirming that once a matter is decided with finality the tribunal cannot revisit it, the court affirmed the importance of finality in arbitration and delineated clear boundaries for judicial intervention under section 34(2)(a)(iii) of the Commercial Arbitration Act 2012 (WA) (“the Act”).

The facts of the case and the decision of the court are provided above ([2025] C.L.J. 39). It is important to note, however, that it was the parties themselves that agreed to bifurcate the proceedings into a liability phase and a quantum phase. In this context, at the heart of the court’s decision lies the doctrine of functus officio, which means that one’s authority ends once they have fulfilled their office. In a non-bifurcated case, once the arbitrators have finally determined all issues before them and issued a final award, their mandate ends (they become functus officio) and they cannot revisit any of the issues discussed in that award. This principle ensures finality: once an arbitrator has fully exercised the entrusted mandate, they cannot reopen the matter unless authorised by the arbitration agreement, legislation or a subsequent agreement of the parties.

The question before the court was whether the same principle applies to a partial award in bifurcated proceedings. Specifically, does the tribunal’s later reconsideration of an issue already decided in a first award constitute a decision made in excess of the tribunal’s jurisdiction? In a 5:2 majority decision, the court answered both questions in the affirmative. It relied on Fidelitas Shipping Co. Ltd. v V/O Exportchleb [1966] 1 Q.B. 630, where Diplock L.J. observed that once an arbitrator renders a partial award on certain issues, they become functus officio with respect to those matters.

There are six main points of note.

First, an issue of terminology: the first award in this matter was both final and binding on the parties as to the matters it decided. It was named in the proceedings and the court’s decision as an “interim final” award. However, despite the contrary view in Emirates Trading Agency L.L.C. v Sociedade de Fomento Industrial Private Ltd. [2015] EWHC 145 (Comm), the status of such awards is better captured by the term “partial (final) award”. These awards are final as to the issues they decide, but are partial because they do not determine all issues brought before the tribunal.

Second, the important question is whether “the challenge that is made goes to the authority of the tribunal” (at [30]) and “which body (i.e. the court or the arbitral tribunal) determines what has been finally decided by an interim award” (at [83]). In this context, there is a fundamental distinction between jurisdictional issues and what are sometimes called “admissibility” issues. Although the “jurisdiction vs. admissibility” terminology can be debated and may not be fully precise, “jurisdiction” concerns whether the tribunal has the authority to decide a particular matter at all. By contrast, “admissibility” refers to whether a claim or argument can be raised before a tribunal that does have authority over the subject matter. The Court emphasised that the tribunal’s attempt to revisit already-decided liability issues was not a matter of “admissibility” within its existing jurisdiction. Instead, it amounted to a jurisdictional overreach. The distinction between matters within the tribunal’s authority and those falling outside it remains crucial: if a tribunal makes legal or factual errors while acting within its jurisdiction, a court usually cannot review those errors at the post-award stage (unless e.g. they constitute a violation of public policy). However, if the tribunal strays beyond its mandate – resolving issues it no longer had authority to address – the resulting award becomes vulnerable to annulment.

Third, the minority argued that reconsideration of issues already addressed in a partial award does not constitute a jurisdictional error justifying curial intervention. According to this view, the Act and the Model Law strictly limit court interference to exceptional cases (for example, a challenge to a jurisdictional award), reflecting the parties’ choice to have disputes decided by an arbitral tribunal rather than by a court. This approach, however, undermines both certainty and the parties’ agreement to bifurcate proceedings and issue separate awards on separate issues. In bifurcated proceedings, functus officio should not be limited to final awards; a partial award that conclusively determines certain questions is final and binding on those questions and the tribunal no longer has the authority to revisit them. Such an award should be treated the same as a decision on an issue for which the tribunal never had jurisdiction.

Fourth, the minority’s view that a tribunal may revisit issues already resolved in a first award disregards the essential purpose and utility of adopting a bifurcated procedure in the first place. Such a position strips partial (final) awards of their finality, undermining the reliance that parties place on having certain key determinations made at the outset. It ignores the fact that, when aspects of a dispute are finally decided at an early stage, the parties’ litigation strategies and the prospect of settlement can shift dramatically based on that newfound clarity. The minority’s approach – in which the tribunal itself might reopen issues closed in a previous stage – does violence to the settled expectations that bifurcation is meant to serve. As such it is not a pro-arbitration approach; rather, it undermines the efficiency and credibility of arbitration as a whole.

Fifth, whether a tribunal’s violation of the res judicata effect of a prior partial final award (or of another court’s decision) can be subject to judicial review is not confined to the jurisdiction-versus-admissibility dipole. Such violations may also be reviewed under the public policy ground of Article V(2)(b) of the New York Convention of 1958 (UNCITRAL, Guide on the Convention on the Recognition and Enforcement of Foreign Arbitral Awards, art. 5(2)(b), 248) and the corresponding provision in Article 34(2)(b)(ii) of the UNCITRAL Model Law.

Finally, both the majority and the minority agreed that if section 34(2)(a)(iii) of the Act is triggered, it allows for a de novo determination of whether a tribunal lacked authority. Under this standard, a state court need not defer to the tribunal’s own view of its jurisdiction. Nevertheless, the principle of kompetenz-kompetenz permits arbitral tribunals to rule on their own jurisdiction in the first instance.

Ultimately, CBI v Chevron reinforces a foundational premise of arbitration: once the bell of finality has tolled on an issue, it cannot be “un-rung”. The tribunal’s jurisdiction is not elastic and its initial mandate extends only so far. If a party attempts – however indirectly – to enlarge that mandate in a subsequent phase of the same arbitration, the courts will act to preserve the integrity of the process. Such “jurisdictional overrun” is not tolerated and courts will grant relief under section 34(2)(a)(iii) of the Act to annul awards that stray beyond the agreed scope.

The Court’s willingness to set aside the second partial award on this basis underscores that, although arbitration promises minimal judicial interference, it does not sacrifice core principles of procedural fairness and finality. By policing the boundaries of arbitral jurisdiction, courts ensure that the process remains governed by the parties’ agreement and the promises made at the outset of the reference to arbitration.

This decision also highlights the importance of carefully structured bifurcation orders and precise delineation of issues at each phase. While splitting the liability and quantum phases can promote efficiency, it may also invite disputes over the classification of issues. Parties should ensure that procedural orders and terms of reference clearly define the scope of each phase. If a point pertains to liability, it must be raised and resolved during the liability phase. Attempting to repackage a resolved liability issue as a quantum question in the second phase will likely fail and could lead to the annulment of the subsequent award. It is a reminder that, while arbitration promises parties control over process and outcome, that promise is anchored in respecting the tribunal’s limited and final mandate. By affirming the functus officio principle in bifurcated proceedings, awards retain their sanctity, finality and enforceability – values at the heart of modern arbitration law and practice.