ECONOMICS · MARKETS · Capital
Infrastructure Economics
Durban Is Moving More Containers. Why Isn't the Backlog Falling?
Durban Gateway Terminal reported improving weekly container throughput even as industry data pointed to roughly 35,083 outstanding vessel-handling moves. Why can both be true? Rokin Research separates incoming work from the surplus capacity needed to clear the backlog, then models recovery times and the economics of accelerating service. The results are illustrative scenarios, not measured financial losses.
Rokin Research
Key takeaways
The transition’s investable bottlenecks are increasingly physical: connection queues, grid equipment, permitting capacity, and skilled delivery.
1. Higher throughput does not guarantee recovery. If 3,400 new moves arrive daily, handling 3,400 moves only stabilises the backlog. 2. Recovery depends on surplus capacity. Clearing 35,083 outstanding moves in 30 days requires approximately 4,569 effective moves per day under the model's assumptions. 3. Backlog is not the same as economic loss. Estimating vessel-delay costs requires reliable arrival, handling and vessel-level timing data.
A container terminal can move more cargo and still leave ships waiting. That is the problem at Durban Gateway Terminal, the large Pier 2 operation in South Africa's main seaport. It is also why gross throughput, on its own, is a poor measure of recovery.
The South African Association of Freight Forwarders (SAAFF), reporting through its BUSA cargo movement updates, puts outstanding vessel-handling work at approximately 35,083 moves in its late-October snapshot. It also describes handling of around 3,300 moves per day, near a working target of 3,400. The accessible EWC summary reports a 9% week-on-week rise in waterside activity for 19–25 October. That percentage and the daily vessel-handling move figures measure different things: gross weekly container throughput does not identify how much outstanding vessel work has been cleared. [1]
The key question is not how much work the terminal performs. It is how much capacity remains after it has handled new arriving vessel work.
A target can be met without clearing anything
Think of the vessel backlog as unfinished work. Every day brings fresh loading and unloading requirements. Every day, the terminal completes some of that work. The queue shrinks only when completed work exceeds newly arriving work on a consistent measurement basis.
Rokin Research uses the 35,083 reported moves as an opening workload stock and assumes, purely for illustration, that 3,400 new moves arrive each day. This is not an observed vessel-arrival series. Under those assumptions, meeting a 3,400-move handling target does not reduce the inherited backlog at all. It merely prevents it from getting larger.
To eliminate the starting stock in 30 days, the terminal would need to complete incoming work plus roughly 1,169 old moves each day. The effective output requirement becomes approximately 4,569 moves per day. At 4,600, the stock clears in 30 calendar days. At 4,000, it takes 59. These are transparent scenarios, not forecasts of when Durban will actually recover. [2]
Sustained effective handling | Illustrative time to clear 35,083 moves |
|---|---|
3,400 moves/day | No clearance |
4,000 moves/day | 59 calendar days |
4,600 moves/day | 30 calendar days |
5,000 moves/day | 22 calendar days |
Source: Rokin Research model v0.2. Assumes 3,400 new vessel-handling moves/day and constant service, with no changes in backlog definition, weather, vessel mix or diversions. The values are not measured Durban recovery outcomes. [2]
That distinction explains the apparent contradiction in the public reporting: higher weekly container throughput can coexist with severe vessel congestion. A better month of output is not necessarily a month of backlog clearance.
The missing number is incoming work
The model's most important unknown is the amount of vessel work newly joining the queue. Two snapshots help illustrate the problem. An earlier SAAFF/BUSA update, with operational notes to 15 October, put 30,352 moves at anchorage. The update with operational notes to 29 October reported 35,083 outstanding vessel moves, an increase of 4,731 in the published estimates. The exact date and scope of the later underlying stock estimate are not established by the accessible synopsis. Consequently, neither a precise thirteen-day rate of growth nor an arrivals count can be claimed as observed. [1] [3]
For a separate diagnostic, the workbook tests a thirteen-day interval (15–28 October) and holds daily service at 3,300 moves. If the later stock really applied on 28 October, and both stock definitions matched, the change would imply about 3,664 new moves/day. Neither condition is verified; 29 October is the date of the distributor's accessible operational update. Under that expressly conditional diagnostic, service at 4,600 moves/day would clear the illustrative stock in about 38 calendar days, instead of 30. [2]
A recovery plan is therefore only as good as its denominator. More crane output might be useful, but it is the sustainable difference between arrivals and completions that determines the backlog path.
A shorter anchorage queue may conceal work still at berth
There is a second measurement problem. The reported workload may chiefly concern vessels waiting at anchorage. When a ship gets a berth, its planned moves may disappear from the anchorage backlog even though the containers have not been handled yet.
The technical report separates work waiting at anchor from work already in berth service. In a constructed example, fast berth admission lowers the first stock while increasing unfinished work at the berth. Only completed moves reduce their combined total. The example does not claim to reconstruct Durban's actual vessel movements. It shows why counting ships admitted to berth cannot, by itself, demonstrate operational recovery. [2]
The same caution applies to costs. A move of pending work is not a delayed container, and neither is automatically one day of ship time. Without vessel-level timestamps and cargo-work records, the model cannot turn the 35,083 moves into a reliable monetary invoice for carriers, importers or South Africa's economy.
The right investment depends on the actual bottleneck
It is tempting to answer the problem by buying more handling equipment. But usable capacity is a chain. Cranes, yard vehicles, stack space, gate bookings, rail evacuation and receiving terminals must all support the extra moves. If the inland stage is already limiting flow, adding capacity at the quay can leave overall completions unchanged.
The research illustrates this with hypothetical compatible capacity limits, not measurements of DGT's gate or yard. Raising only a non-binding crane stage produces no additional completed work. Coordinating improvements through the binding stage does. The same principle applies to diversions: moving calls to another terminal only helps if it has genuinely available berth, yard and inland capacity. [2]
The technical report also tests the economic trade-off. Recovering faster may reduce unfinished-work exposure, but more intensive operations can cost more. Under an explicitly fictional set of rand-valued cost coefficients, the cost proxy reaches a minimum near 4,724 effective moves/day, slightly above the capacity needed for a 30-day result. That figure is not an operator recommendation, a market quotation or the measured value of Durban's disruption. Its purpose is to show which costs and capacity constraints a credible decision would need to establish. [2]
South Africa's transport minister said in late October that incoming equipment and further improvements could take several months, with progress anticipated through February 2027. This is an attributed expectation, not an observed completion date or a validation of the model's faster cases. [4]
Judge recovery by the work actually removed
Durban's public data support a useful conclusion, but not a precise loss estimate. They show a substantial reported work backlog alongside some improvement in throughput. What remains unverified is the daily balance of new vessel work, completed work and unfinished work split between anchorage and berth.
For port operators and capital allocators, the lesson is practical: track the net clearance margin and invest where additional capacity becomes completed work. A productivity target is not a recovery target if it merely keeps pace with arrivals. And a shorter line of anchored ships is not necessarily proof that the port has finished the underlying cargo work.
Research note. This article combines dated SAAFF/BUSA observations with a constructed stock-flow and economic sensitivity model. The 30-day, 38-day and cost-optimal examples depend on hypothetical arrivals, snapshot timing and operating/cost inputs. They do not represent observed vessel delays, actual DGT operating commitments or audited financial losses. [1] [2] [3]
Article references
[1] SAAFF/BUSA via EWC. Cargo Movement Update No.305 – DGT update as of 29 October (25 Oct 2026 report; EWC posted 29 Oct 2026). Open source. Directly accessible distributor synopsis: 35,083 moves, ~3,300/day, target 3,400 and +9% waterside growth; does not display exact 31,149 TEU.
[2] Rokin Research. Durban Gateway Terminal: Backlog Recovery, technical report and editable model v0.2 (30 Oct 2026). Technical report and Excel model in the accompanying package. Own derived scenarios and equations. No public URL assigned; see supplied PDF/XLSX.
[3] SAAFF/BUSA via EWC. Cargo Movement Update No.303 – DGT update as of 15 October (11 Oct 2026 report; EWC posted 15 Oct 2026). Open source. Accessible distributor synopsis for the 30,352-move anchorage stock; original PDF not independently read.
[4] Karen Singh, The Mercury / IOL. Durban Pier 2 backlog could take until February to clear, says Creecy (30 Oct 2026). Open source. Attribution of minister’s expectation; not proof of future clearance.