FGDI, LLC operates a grain elevator at the Alabama State Docks in Mobile, Alabama. On December 7, 2002, M/V LORE-LAY was berthed in the Port of Mobile and suffered a crack in the bulkhead between her fuel oil tank and ballast tank, resulting in an oil spill. Thereafter, the Coast Guard closed FGDI’s loading berth until the area could be cleaned. FGDI and MTV LORELAY have at all times agreed that M/V LORELAY is liable under the Federal Oil Pollution Act of 1990 to pay FGDI for damages resulting from the oil spill and the delay the spill caused FGDI’s business. However, the parties disagreed as to the duration of the delay and the amount of the resulting damages and thus proceeded to a bench trial. The district court identified six different kinds of damage sustained and awarded a total of $167,173.33 to FGDI. M/V LORELAY appeals the judgment. FGDI cross-appeals.

M/V LORELAY contends that the factual findings on which the damage award depends are clearly erroneous and further contends that the district court awarded damages that had not been sufficiently proven by FGDI. M/V LORELAY contends that each component of the district court’s award is affected by these errors and, therefore, the entire judgment must be vacated and the case remanded for reconsideration by the district court. FGDI answers that the district court’s factual findings are not in error, and with one exception (the subject of its cross-appeal), fully support the damage award. FGDI seeks an increase in the amount of the damages awarded as compensation for costs resulting from delay to railcars on the CN and BNSF rail lines, arguing that the court’s factual findings support an award twice that ordered for that damage component. FGDI also contends that it proved all its damages to a reasonable certainty, a standard that the parties agree applies.

After a thorough review of the briefs and the record in this case, we hold that the district court’s judgment is premised upon clearly erroneous factual findings and depends, in part, on inadequate proof of damages.

*856 VESSEL DELAY

The parties agree that the loading of three vessels, ANGELIC PEACE, TAMPICO ALTO, and PRINCESS VANYA, was affected by the closure of the loading berth. The district court found that ANGELIC PEACE was delayed four days, that TAMPICO ALTO was not delayed, and that PRINCESS VANYA was delayed four days. These findings of fact are unsupported by the testimony or documentary evidence and rely on speculation. They are, therefore, clearly erroneous.

The key testimony in this case was provided by FGDI’s President and CEO, Steven Speck. Speck testified that the general practice at FGDI was that the first vessel to pass the required cleanliness inspections was the first vessel to go to the loading berth. (R.l-40 at 82, 88.) He also testified that this general practice would have been followed in this particular instance. (R.l-40 at 34) (“Q: All things considered, if there had been no spill, was it your plan to load the TAMPICO or another vessel first? A: We would have— we would have loaded whoever came in and passed their inspections first.”) Speck testified that one of the reasons that TAMPICO ALTO actually loaded first was that she was the first to pass inspection. (R.l-40 at 89.)

It is undisputed that no other vessel was ready to load at 1000 December 13, when TAMPICO ALTO was ordered to move into the loading berth. TAMPICO ALTO was the first vessel to pass inspection and the first vessel to go to the loading berth; ANGELIC PEACE was the second vessel to pass inspection and the second to go to the loading berth; PRINCESS VANYA was the last of the three vessels to pass inspection and to go to the loading berth. Thus, the oil spill did not affect the order in which the vessels loaded. 1

*857 The delay each vessel experienced as a result of the spill must be calculated by counting the time between when the vessel was ready and able to move into the loading berth (but was prevented from doing so as a result of the closure of the loading berth) and when the vessel actually moved to the loading berth. TAMPICO ALTO was ready to move to the loading berth at 0920 December 11 but was unable to do so because the loading berth was closed due to the oil spill. TAMPICO ALTO actually began shifting to the loading berth at 1130 December 13. Thus, TAMPICO ALTO was delayed just over two days as a result of the oil spill. 2

M/V LORELAY argues that once TAMPICO ALTO was in the loading berth, any time that the subsequent vessels waited is not delay attributable to the oil spill but rather an inescapable result of the fact that FGDI’s loading berth could accommodate only one ship at a time. This argument fails to account for the domino effect that delay in loading the first ship had on subsequent ships. If no spill had occurred and TAMPICO ALTO had been able to enter the loading berth on December 11, it would have finished loading late on December 12. Then, when ANGELIC PEACE passed inspection at 1230 December 13, it would have been able to proceed to the loading berth immediately. Instead, ANGELIC PEACE waited for TAMPICO ALTO to complete loading and exit the loading berth, actually moving to the loading berth beginning at 0505 December 15. Thus, as a result of the oil spill, ANGELIC PEACE was delayed approximately one day and 18 hours.

Finally, PRINCESS VANYA passed inspection at 1230 December 15, but she could not move to the loading berth until ANGELIC PEACE had cleared the loading berth. It actually took ANGELIC PEACE 3 days and 20 hours to move to the loading berth and load. Thus, had ANGELIC PEACE moved to the loading berth at 1230 December 13, it would be reasonable to expect that she would have been ready to leave the loading berth at approximately 0830 December 17. PRINCESS VANYA could have proceeded to the loading berth at 0830 December 17, 1 day and 17 hours before she actually moved to the loading berth at 0130 December 19. Thus, PRINCESS VANYA was delayed 1 day and 17 hours as a result of the oil spill.

DAMAGES RESULTING FROM VESSEL DELAY

Vessel Dispatch and Demurrage

The vessel delay times must be used to calculate the vessel dispatch and demurrage component of the damage award. According to Speck’s testimony, dispatch is a payment made to the grain elevator if the elevator loads the vessel in less time than agreed upon in advance. (R.l-40 at 103.) FGDI’s claim for lost dispatch is a claim for lost profits result *858 ing from delay in loading attributable to the spill. FGDI claimed one day lost dispatch on ANGELIC PEACE and one day lost dispatch on PRINCESS VANYA. Demurrage is a penalty that the grain elevator pays the vessel if the grain elevator fails to load the vessel in the agreed upon time. (R.l-40 at 104.) FGDI’s claim for demurrage is a claim that it became liable for payment to the vessel as a result of the delay caused by the oil spill. FGDI claimed two days’ demurrage on TAMPICO ALTO only. Relying on its determination that TAMPICO ALTO was not delayed, the district court denied any dispatch or demurrage for TAMPICO ALTO. The district court’s order states that it awarded one day demurrage for both ANGELIC PEACE and PRINCESS VANYA, but the $4,750 awarded for ANGELIC PEACE and $6,250 awarded for PRINCESS VANYA are actually daily dispatch rates, not demurrage rates. (R.l-40 at 41-42.) On remand, the district court shall reconsider the dispatch and demurrage awards in light of the above discussion regarding vessel delay.

Interest

The interest award relies on the erroneous factual findings regarding the duration of vessel delay. Therefore, the interest award must also be revisited in light of the above discussion of vessel delay.

Overtime

The district court found that, as a result of the oil spill, FGDI was required to use three shifts of overtime to load ANGELIC PEACE. This finding is unsupported by the evidence and therefore clearly erroneous. The testimony at trial was that, prior to the oil spill, FGDI expected to load ANGELIC PEACE in approximately four days. (R.l-40 at 96.) Speck did not know how many overtime shifts would have been required if ANGELIC PEACE had been loaded as originally planned. (R.l-40 at 97.) But (using five overtime shifts) the vessel was actually loaded in 3 days 20 hours, only 4 hours short of the estimated four days. So, significant overtime would have been necessary to load the ship in the amount of time FGDI expected, even if no oil spill had occurred.

On remand, the district court shall reconsider and, if necessary, recalculate the damages for overtime in light of the above discussions of vessel delay and actual time necessary to load ANGELIC PEACE and taking into account that the delay caused by the spill might (or might not) have caused FGDI to pay for weekend and evening work rather than weekday work. The damages attributable to overtime are the difference between the cost of the labor that would have been necessary to load ANGELIC PEACE on December 13 through December 17 and the actual cost of the labor used to load ANGELIC PEACE (on December 15 through December 19). 3

Delay and Holding Charges for Railcars on the CN and BNSF Lines

Both parties appeal the district court’s award of $98,005 for railcar delay. *859 Both contend that the award rests upon a clearly erroneous finding of fact that the oil spill caused approximately one-half of the railcar delay. M/V LORELAY argues that the spill delayed operations at the loading berth by less than two days (the time that TAMPICO ALTO was delayed) and thus delayed the railcars for less than two days. FGDI contends that MTV LO-RELAY must be responsible for the costs of holding all railcars from December 7 (the date of the oil spill) until those rail-cars were restarted.

The factual finding that the oil spill caused one-half of the railcar delay (and therefore holding charges) is unsupported by the evidence and therefore clearly erroneous. The record demonstrates that most of the stopped railcars were carrying soybeans and could not be moved until ANGELIC PEACE had been loaded with soybeans. (R.l-40 at 36-37, 53.) Thus, if there had been no oil spill and ANGELIC PEACE had been able to load as soon as she was ready, the railcars still would have been delayed from the date they were stopped until ANGELIC PEACE was loaded. On remand, the district court shall recalculate the delay to rail cars on the CN and BNSF lines caused by the oil spill by determining how much longer the rail cars were delayed than they would have been had ANGELIC PEACE been able to move into the loading berth when she was ready to do so, at 1230 December 13.

Holding Charges for FGDI Leased Railcars and Alabama State Docks

FGDI claimed damages resulting from the delay in emptying and moving railcars that were leased to FGDI and in a holding area in Mobile. (R.1^40 at 27-28.) These railcars contained primarily wheat, and their contents were loaded on TAMPICO ALTO. (R.l-40 at 27-29, 32.) The district court found that FGDI was entitled to four days’ worth of delay charges on these cars. This finding is clearly erroneous as it is unsupported by the evidence. 4 On remand, the district court shall consider the above discussion regarding delay to the vessels, awarding FGDI damages for the costs it incurred as a result of the delay in moving the railcars in the holding area in Mobile that was caused by the delay in loading TAMPICO ALTO.

For the above reasons, the judgment of the district court is vacated. The case is remanded to the district court for reconsideration and recalculation of damages consistent with this opinion.

VACATED AND REMANDED.