Slicing the Award Too Thin: HPCL v. G.R. Engineering and the Drift from Severance to Modification under Section 34

Introduction

Over the last decade, India has sought to establish itself as an arbitration-friendly jurisdiction through the 2015 and 2019 amendments to the Arbitration and Conciliation Act, 1996 (“the Act”) and institutional initiatives such as the New Delhi International Arbitration Centre. However, judicial practice continues to undermine this legislative intent. This article proceeds on the argument that Section 34, particularly its grounds under Section 34(2)(b)(ii) and Section 34(2A), was designed as a narrow safeguard- a supervisory mechanism rather than an appellate one, as recognised in McDermott International v. Burn Standard Co. Ltd. and reaffirmed in Ssangyong Engineering & Construction Co. Ltd. v. NHAI (hereinafter “Ssangyong Engineering”).  Yet, courts have transformed it into a site of expansive intervention, blurring the boundary between annulment and modification.

The Bombay High Court’s decision in Hindustan Petroleum Corporation Ltd. v. G.R. Engineering Pvt. Ltd. (hereinafter “HPCL”) arose out of a dispute concerning delays in the execution of a turnkey contract for a petroleum project. HPCL had withheld approximately ₹5.83 crores as liquidated damages on account of the delay, while the arbitral tribunal directed a refund, holding that HPCL had failed to establish actual loss. HPCL challenged this portion of the award under Section 34, alleging that the tribunal ignored the principles under Section 74 of the Contract Act and binding precedent. The judgment exemplifies this shift- although the Court purported to sever only the liquidated-damages component, its reasoning effectively rewrote the award’s operative effect, collapsing the line between annulment and modification. The Court’s decision to annul only the liquidated damages component while directing that the issue be reopened for re-arbitration effectively altered the operative outcome of the award. In substance, this amounted to a judicial re-determination of the dispute’s core issue, an action that Section 34 does not authorise, thus blurring the procedural distinction between setting aside and modifying an award. Its significance lies not merely in the outcome but in its doctrinal posture. HPCL is among the first post-amendment cases where a High Court explicitly employed the doctrine of severability to partially set aside an arbitral award while simultaneously reopening the issue for fresh arbitration. This marks a departure from the principle that Section 34 permits annulment, not modification, thereby positioning HPCL as a crucial turning point in understanding the evolving scope of judicial intervention in India’s arbitration regime.

Some have long criticised judicial overreach under Section 34, particularly post-Saw Pipes and Western GECO. Yet, the novelty of HPCL lies in its use of severability itself as the instrument of intervention, a shift largely unexplored in existing literature.

This article argues that such an approach stretches Section 34 beyond its remit. It proceeds in five parts- Part I traces the origins of severability; Part II analyses the HPCL ruling; Part III situates it within recent trends, including Gayatri Balasamy v. ISG Novasoft Technologies Ltd. (hereinafter “Gayatri Balasamy”), and State of Rajasthan v. Sanwariya Infrastructure Pvt. Ltd. (hereinafter “Sanwariya Infrastructure”); Part IV considers counterarguments; and Part V offers a comparative perspective on the principle of minimal intervention.

From Preserver to Modifier

The doctrine of severability in Indian arbitration was conceived as a tool of restraint. Section 34 permits annulment of awards only on narrow grounds, such as procedural defects under Section 34(2)(a) or substantive violations under Section 34(2)(b)(ii) (‘public policy’) and Section 34(2A) (‘patent illegality’ in domestic awards). Severability was intended to work within this limited framework, allowing courts merely to excise the defective portion while preserving the rest. It is significant that Section 34 itself contains no reference to partial annulment. Severability is a judicially implied doctrine, originally meant to avoid collateral damage to valid findings, but its extension into redrafting outcomes lacks textual basis.

Early jurisprudence reflected this design. In McDermott International v. Burn Standard, the Supreme Court held that Section 34 is not appellate. In Associate Builders v. DDA (hereinafter “Associate Builders”), perversity was confined to findings based on no evidence, irrelevant evidence, or omission of vital evidence. In Ssangyong Engineering & Construction Co. Ltd. v. NHAI”), the Court clarified that mere interpretive disagreement cannot justify intervention.

Over time, however, this doctrine, once conceived as a shield of restraint, has been recast into a sword of intervention. ONGC v. Saw Pipes, broadened “public policy,” ONGC v. Western GECO expanded “fundamental policy,” and High Courts such as in Sanwariya Infrastructure have used severability to reappraise causation and damages. Against this backdrop, HPCL represents the clearest example of severability enabling judicial modification rather than restraint.

HPCL and the Doctrinal Drift Towards Judicial Editing

The Bombay High Court’s judgment in HPCL illustrates the evolving use of severability under Section 34. At first sight, the Court’s decision to annul only the liquidated damages (“LDs”) portion of the award suggests judicial restraint. On closer analysis, however, the reasoning demonstrates how severability is being repurposed into a tool of judicial modification.

• The Tribunal’s Findings on LDs

The arbitral tribunal found that HPCL was not entitled to withhold liquidated damages and directed refund of approximately ₹5.83 crores. The tribunal reasoned that while a delay may entail some loss of investment, HPCL had failed to substantiate such loss, despite being in a position to do so (¶30-35 of the award, summarised at ¶26-31 of the judgment). Notably, the tribunal concluded that HPCL had not produced adequate proof of actual loss, but it did not expressly analyse whether the stipulated damages clause, prescribing 0.5% per week capped at 5%, represented a genuine pre-estimate of loss or whether the loss was “difficult or impossible to prove.”

Admittedly the tribunal did not use the words “difficult or impossible to prove” (HPCL ¶44), yet it explicitly found that HPCL had failed to prove loss despite being in a position to do so (tribunal ¶30-35; judgment summary ¶26-31). This finding implies that losses were not inherently unprovable; merely unproven on the evidence present. Under Ssangyong and Associate Builders, a debatable factual conclusion or a thinly reasoned finding is not the same as ‘patent illegality’ under Section 34(2A); absence of depth does not automatically elevate to perversity.

This position aligns with Dyna Technologies Pvt. Ltd. v. Crompton Greaves Ltd., where the Supreme Court clarified that insufficiency of reasoning, so long as the arbitrator’s path of reasoning is discernible, does not justify setting aside an award under Section 34.

• The High Court’s Critique

The High Court seized on this omission. Drawing on Kailash Nath v. DDA (hereinafter “Kailash Nath”), it observed that under Section 74 of the Indian Contract Act, liquidated damages may be awarded without proof of actual loss only where the loss is inherently difficult or impossible to prove (¶42-44). The Court held that the tribunal had failed to undertake this inquiry, had ignored authorities cited by HPCL, and had offered a reasoning “devoid of analysis” (¶33-36, ¶44). On that basis, the Court declared the LD portion of the award “manifestly arbitrary, perverse and contrary to the fundamental policy of Indian law” (¶46), thereby invoking the ground under Section 34(2)(b)(ii), as well as stretching the scope of Section 34(2A).

The High Court’s reasoning deserves fuller engagement. The Court’s analysis hinged on whether the award was “sparsely reasoned” or “devoid of reasoning,” a distinction derived from Kailash Nath v. DDA. However, this article adopts the standard articulated in Associate Builders v. DDA, where an award is considered devoid of reasoning only if it offers no discernible link between evidence and conclusion. Sparse reasoning, even if unelaborated, remains within arbitral discretion. On this view, the arbitral tribunal in HPCL was not devoid of reasoning – it addressed the evidentiary question of loss, albeit briefly, and thus met the minimal standard of rational explanation.

• Conflating Legal Error with Interpretive Disagreement

This analytical gap in the Court’s application of severability is compounded by a second doctrinal drift, the conflation of legal error with interpretive disagreement. The Court’s reliance on perversity further illustrates the drift. The tribunal’s findings were based on contemporaneous project progress reports, correspondence exchanged between HPCL and G.R. Engineering, and delay analyses forming part of the contractual record. These materials led the tribunal to conclude that HPCL was in a position to substantiate its loss but failed to do so. This evidentiary evaluation, though succinct, demonstrates that the tribunal did engage with the material on record. Its reasoning may have been terse, but it cannot be equated with a complete disregard of evidence.

By treating this limited articulation as perversity, the High Court effectively blurred the line between a thinly reasoned award and one that is devoid of reasoning, thereby expanding the threshold for judicial interference under Section 34(2A). The Supreme Court in Ssangyong Engineering explicitly cautioned that interpretive disagreement does not amount to patent illegality. Yet in HPCL, the Court converted what may at best have been a misapplication of Kailash Nath into perversity, thereby exercising appellate-style oversight under Section 34.

• Severability Without Separability

Having invalidated the LD portion, the Court applied the doctrine of severability. It clarified in paragraph 48, in addition to the holding that partial annulment would have no bearing on the other claims, that-

“Nothing in the component of the Impugned Award dealing with the challenge to the retention of liquidated damages that is being set aside in this judgement is interlinked and interconnected with the rest of the Impugned Award.”

The Court’s assumption that the liquidated damages component was unconnected to the rest of the award is unconvincing. The finding on liquidated damages stemmed directly from the tribunal’s assessment of delay, causation, and performance- issues that also underpinned HPCL’s claims for payment, extension of time, and completion. By isolating this portion without re-examining those shared factual foundations, the Court treated as severable what was, in fact, conceptually and evidentially intertwined. The supposed “independence” of the LD component is therefore illusory, and its excision disturbs the award’s internal coherence.

This holding fails the substratum test articulated by the Constitution Bench in Gayatri Balasamy: severance is permitted only where the impugned portion is-

• factually independent,
• legally separable and,
• clearly shown not to affect the award’s substratum (Gayatri Balasamy ¶33–36; concurring ¶142–152).

On the facts, the LD finding was derived from the tribunal’s assessment of delay, causation and performance (tribunal ¶30–35; judgment ¶26–31). The High Court does not explain why, despite its criticism of causation (¶45), it later concluded LDs were non-interconnected (¶48). Proper application of the substratum test would require a paragraph-by-paragraph analysis showing how excising paragraph 7 leaves the substratum intact, an analysis absent from HPCL. This level of scrutiny is implicit in Balasamy, where the Court emphasised that severance is permissible only when the impugned portion is factually and legally independent and its removal does not disturb the award’s foundation- an inquiry that necessarily demands granular examination of the reasoning structure.

Finally, the Court’s remedial order underscores its slippage from annulment to modification. In ¶46, it quashed paragraph 7 of the award as perverse and contrary to fundamental policy. However, in ¶69[b] it went further, declaring that ‘the arbitration agreement would continue insofar as it relates to liquidated damages,’ thus authorising a re-arbitration of the issue. The Court’s order (¶69[b]) effectively reopens the dispute: quashing paragraph 7 but preserving the balance of the award and allowing re-arbitration converts annulment into practical re-determination. Section 34’s remedial architecture permits annulment (Section 34(2)) or remission to the tribunal to cure defects (Section 34(4)); it does not empower courts to excise a component and then invite a second arbitral proceeding to replace the excised part. This functional modification undermines finality and the statutory division between annulment and merits re-determination.

Section 34 empowers only annulment under Section 34(2) or remission under Section 34(4); it does not contemplate reopening claims for fresh arbitration. By doing so, the Court effectively modified the operative effect of the award, a remedy the Supreme Court has rejected as in NHAI v. M. Hakeem.

HPCL in Context- An Emerging Judicial Trend

Recent cases reinforce this trajectory. In NHAI v. M. Hakeem, the Supreme Court held in categorical terms that Section 34 does not permit modification, only annulment. However, this rigid position was subsequently qualified in Gayatri Balasamy, where the Court recognised that limited partial annulment may be permissible if the impugned portion of the award is demonstrably severable, legally independent, and does not disturb the substratum of the award. Thus, while Hakeem affirmed the binary nature of Section 34 remedies, Balasamy carved out a narrow doctrinal exception that HPCL appears to have stretched beyond its intended scope. Yet subsequent High Court rulings, including Sanwariya Infrastructure and Delhi Airport Metro Express v. DMRC, have invoked severability to adjust damages or causation, blurring annulment and redrafting.

Notably, other High Courts have deployed severability in ways that blur the line between annulment and modification. In Sanwariya Infrastructure, for instance, the Court annulled only the portion of the award imposing delay penalties. It is important to note that Sanwariya Infrastructure was decided after Gayatri Balasamy. Despite the Supreme Court’s clear articulation in Balasamy that partial annulment must satisfy the substratum test, requiring that the impugned portion be factually and legally independent, the Rajasthan High Court did not apply this standard. Its failure to do so reflects a post-Balasamy regression, where High Courts invoke severability in form but not in method. On its face, this appeared restrained. However, the Court reassessed causation and quantum matters squarely within arbitral discretion. Consequently, severability functioned less as a shield of preservation and more as a vehicle for judicial substitution.

By contrast, the court in Gayatri Balasamy sought to discipline severability. It held that partial annulment is permissible only when the impugned portion is independent, severable, and does not disturb the substratum of the award (¶33-36; concurring ¶142-152). Yet both Sanwariya and HPCL invoked severability without applying this rigorous test.

Moreover, this drift runs counter to the Supreme Court’s recalibrations of Section 34(2)(b)(ii) (‘public policy’) and Section 34(2A) (‘patent illegality’) in Associate Builders and Ssangyong Engineering, which confined perversity strictly to findings based on no evidence, irrelevant evidence, or omission of vital evidence. Thus, what began as a doctrine of restraint is increasingly repurposed as a tool of judicial editing, undermining arbitral autonomy.

Counterarguments and Responses

Any critique of HPCL must directly confront the High Court’s justifications. Two stand out –  first, that the tribunal failed to apply the standard in Kailash Nath, and second, that it ignored vital evidence, thereby rendering its findings perverse. At first glance, these arguments appear persuasive; however, on closer examination, they collapse. As Ssangyong Engineering (¶34) clarified, a tribunal retains a “right to be wrong” and mere legal error is not a ground for annulment.

The Court’s primary reliance was on Kailash Nath. It reasoned that under Section 74 of the Contract Act, liquidated damages may be awarded without proof of actual loss only when the loss is “difficult or impossible to prove” (¶44). Because the tribunal did not expressly make this determination, the Court characterised its reasoning as contrary to the fundamental policy of Indian law (¶46). Yet this critique overlooks the tribunal’s own findings – it held that HPCL was in a position to prove its loss but failed to do so (¶30-35). Implicitly, then, the tribunal concluded that the loss was not impossible to prove. Even if this was a debatable application of Kailash Nath, it was not an omission so grave as to constitute perversity. The Supreme Court has consistently held, most recently in McDermott, Ssangyong and Dyna Technologies– that mere legal error or misapplication of law does not warrant judicial intervention under Section 34. The High Court’s reliance on Kailash Nath ignores the difference between a legal misapplication and an evidentiary omission. The tribunal’s express factual finding that HPCL failed to prove loss (tribunal ¶30–35) is not the same as a refusal to apply the Kailash Nath test; it is an evidentiary assessment. Were the tribunal’s conclusions demonstrably unsupported by any evidence, annulment would be appropriate; but Associate Builders confines ‘perversity’ to findings based on ‘no evidence’ or irrelevant material (¶31). HPCL treats arguable legal error as perversity, thereby expanding Section 34(2A) beyond its doctrinal limits.

The High Court’s further suggestion that the tribunal’s omission to cite Kailash Nath amounted to perversity is equally misplaced. Failure to engage with precedent is a legal error, not an evidentiary omission. Section 34(2A) does not allow courts to correct legal errors; if it did, every misapplication of precedent would be rebranded as perversity, transforming Section 34 into a disguised appellate forum.

Equally unconvincing is the claim that the tribunal ignored vital evidence. The award engaged with delay reports, contractual provisions, and correspondence (¶26-31), albeit briefly. As the Supreme Court explained in Associate Builders (¶31), perversity arises only when an award is based on no evidence or when vital evidence is altogether disregarded. Sparse reasoning does not satisfy this threshold.

Consequently, even if the tribunal erred in applying Section 74, the High Court’s response, excising the LD portion and permitting re-arbitration (¶69[b]), amounted to modification, a remedy that finds no footing in Section 34. The statute authorises only annulment under Section 34(2) or remission to the tribunal under Section 34(4); modification lies outside its text.

Minimal Judicial Intervention

At the heart of the Arbitration and Conciliation Act, 1996 lies the principle of minimal judicial intervention. Section 5 provides that “no judicial authority shall intervene except where so provided in this Part,” underscoring the legislature’s intent to limit curial interference. Section 34, carefully structured through sub-sections 34(2), 34(2A) and 34(4), provides for procedural and substantive safeguards, patent illegality in domestic awards, and remission to eliminate grounds for annulment respectively, was designed as a narrow safety valve, permitting annulment only on exceptional grounds. The Supreme Court has consistently affirmed this principle. In McDermott International v. Burn Standard, it stressed that Section 34 does not confer appellate powers. In Ssangyong Engineering (¶34), it warned that interpretive disagreements cannot be recast as patent illegality.

By contrast, the Bombay High Court in HPCL set aside a portion of the award because the tribunal’s reasoning was thin and failed to discuss certain authorities (¶33-36, ¶44-46). This is appellate-style scrutiny, alien to the UNCITRAL Model Law and to the standards observed in pro-arbitration jurisdictions.

The cumulative effect of such judicial approaches is a perceptible erosion of the legislative ideal of minimal curial intervention. By recasting Section 34 into a vehicle for merits review through doctrines like severability and perversity, Indian courts risk transforming what was meant to be a narrow supervisory power into an appellate mechanism in disguise.

To arrest this drift, the Supreme Court must restate the law along three axes-

1. The substratum test- Severability should only apply when the impugned portion is demonstrably independent and its removal does not affect the award’s foundation, as clarified in Gayatri Balasamy (¶33-36, ¶142-152).
2. Narrow conception of perversity- Perversity should remain confined to instances where findings are based on no evidence, irrelevant evidence, or where vital evidence has been ignored (Associate Builders, ¶31). Sparse reasoning alone should not suffice.
3. No disguised modification- Courts must recognise that modification, whether explicit or cloaked as severance, lies beyond the remit of Section 34. Annulment is a binary choice- either set aside the award, or uphold it.

Conclusion

The decision in HPCL v. G.R. Engineering marks a deeper drift in India’s arbitration jurisprudence. Section 34, conceived as a narrow supervisory mechanism, is increasingly applied as an appellate tool under the garb of “severability” and “perversity”. By treating sparse reasoning as perversity and reopening issues for re-arbitration, the Court blurred the line between annulment and modification, undermining the autonomy that arbitration is meant to protect.

To restore coherence, courts must apply the substratum test rigorously, limit perversity to findings wholly unsupported by evidence, and avoid any form of disguised modification. Clarifying these limits would harmonise Indian practice with the UNCITRAL Model Law and reaffirm the legislative intent of minimal intervention. Ultimately, the challenge lies not in the statute but in its interpretation, unless judicial oversight is confined to its proper bounds, India’s aspiration to be a credible arbitration hub will remain compromised by its own jurisprudential excess.

About the Author

Anubhuti Raje is a final-year B.A. LL.B. (Hons.) student at Gujarat National Law University, Gandhinagar with interests spanning to dispute resolution, white-collar crime, and public law.